The Board of Directors of the Radiological Society of North America and theSociety’s Refresher Course Committee acknowledge with warm appreciation

the scientific contributions ofFredrick A. Mann, MD,

in his role as editor of this syllabus.

2004 Syllabus

Emergency RadiologyCategorical Course in Diagnostic Radiology

EDITOR

Frederick A. Mann, MDSEATTLE, WASH

Presented at the 90th Scientific Assembly and Annual Meetingof the Radiological Society of North America

November 28–December 3, 2004

Robert R. Hattery, MDChairman

Board of Directors

Robert A. Novelline, MDChairman

Refresher Course Committee

The Radiological Society of North America is a nonprofit organization.

The Radiological Society of North America subsidizes the production ofcategorical and special course syllabi. Financial losses incurred in syllabiproduction are absorbed by the RSNA. Any net revenues from the sale ofsyllabi are transferred to the RSNA Research and Education Foundation.

©2004 by the Radiological Society of North America, Inc820 Jorie Blvd, Oak Brook, IL 60523-2251

3

Contents

5 Preface

Frederick A. Mann, MD

7 The Emergence of Emergency Radiology

Robert A. Novelline, MD

11 Nontraumatic Neurologic Emergencies

A. Gregory Sorensen, MD

17 Traumatic Brain Injury: ImagingUpdate 2004

Alisa D. Gean, MD, Christine Glastonbury,MBBS, and D. Christian Sonne, MD

33 CT for Thromboembolic Disease:Protocols, Interpretation, and Pitfalls

Lacey Washington, MD

47 CT of Nontraumatic Aortic Emergencies

O. Clark West, MD, and Sanjeev Bhalla, MD

59 Cardiac Applications for Multi–DetectorRow CT in the Emergency Department

Udo Hoffmann, MD, Ricardo C. Cury, MD,Maros Ferencik, MD, PhD, Fabian Moselewski, BS,Suhny Abbara, MD, and Thomas J. Brady, MD

71 Imaging of Blunt Chest Trauma

Nisa Thoongsuwan, MD, Jeffrey P. Kanne, MD,and Eric J. Stern, MD

81 Imaging Diagnosis of Thoracic Aorta andGreat Vessel Injuries

Stuart E. Mirvis, MD

91 CT of Abdominal Trauma: Part I

James T. Rhea, MD

101 CT of Abdominal Trauma: Part II

Kathirkamanathan Shanmuganathan, MD

113 Imaging the Pediatric Patient with AcuteAbdominal Disease

Carlos J. Sivit, MD

119 The Role of CT in Acute AbdominalDisease: Pitfalls and Their Lessons

Dean D. T. Maglinte, MD, James T. Rhea, MD,and M. Stephen Ledbetter, MD

133 The Contemporary Role of ConventionalRadiographs in Evaluating the AcuteAbdomen

Stephen R. Baker, MD

143 Imaging of Cervical Spine Trauma

C. Craig Blackmore, MD, MPH

151 Imaging Spine Trauma in the Elderly

Friedrich M. Lomoschitz, MD, C. Craig Blackmore,MD, MPH, and Frederick A. Mann, MD

159 Imaging of Thoracolumbar Spine Trauma

Georges Y. El-Khoury, MD

169 Imaging of Upper Extremity Injuriesin Children

Diego Jaramillo, MD, MPH

175 High-Energy Blunt-Force Injuries to theUpper Extremity

Thurman Gillespy III, MD

187 Imaging Low-Energy Upper ExtremityInjuries

Viktor M. Metz, MD, and Marcel O. Philipp, MD

197 Low-Energy Injuries of the Lower Limb

Philip M. Hughes, MBBS, MRCP, FRCP

217 Pediatric Lower Extremity Trauma

Susan D. John, MD

4

ContributorsSuhny Abbara, MDDepartment of RadiologyMassachusetts General HospitalBoston, Mass

Stephen R. Baker, MDDepartment of RadiologyUniversity HospitalNew Jersey Medical SchoolNewark, NJ

Sanjeev Bhalla, MDMallinckrodt Institute of RadiologyWashington University School of

MedicineSt Louis, Mo

C. Craig Blackmore, MD, MPHDepartment of RadiologyHarborview Medical CenterUniversity of WashingtonSeattle, Wash

Thomas J. Brady, MDDepartment of RadiologyMassachusetts General HospitalBoston, Mass

Ricardo C. Cury, MDDepartment of RadiologyMassachusetts General HospitalBoston, Mass

Georges Y. El-Khoury, MDDepartment of RadiologyUniversity of Iowa Hospitals and

ClinicsIowa City, Iowa

Maros Ferencik, MD, PhDDepartment of RadiologyMassachusetts General HospitalBoston, Mass

Alisa D. Gean, MDDepartment of RadiologyUniversity of California, San FranciscoSan Francisco, Calif

Thurman Gillespy III, MDDepartment of RadiologyUniversity of WashingtonHarborview Medical CenterSeattle, Wash

Christine Glastonbury, MBBSDepartment of RadiologyUniversity of California, San FranciscoSan Francisco, Calif

Udo Hoffmann, MDDepartment of RadiologyMassachusetts General HospitalBoston, Mass

Philip M. Hughes, MBBS,MRCP, FRCP

Department of RadiologyDerriford HospitalPlymouth University HospitalsPlymouth, England

Diego Jaramillo, MD, MPHDepartment of RadiologyChildren’s Hospital of PhiladelphiaPhiladelphia, Pa

Susan D. John, MDDepartment of RadiologyUniversity of Texas Houston

Medical SchoolHouston, Tex

Jeffrey P. Kanne, MDDepartment of RadiologyHarborview Medical CenterUniversity of WashingtonSeattle, Wash

M. Stephen Ledbetter, MDDepartment of RadiologyBrigham and Women’s HospitalBoston, Mass

Friedrich M. Lomoschitz, MDDepartment of RadiologyVienna Medical SchoolUniversity of ViennaVienna, Austria

Dean D. T. Maglinte, MDDepartment of RadiologyIndiana University School of MedicineIndianapolis, Ind

Frederick A. Mann, MDDepartment of RadiologyHarborview Medical CenterSeattle, Wash

Viktor M. Metz, MDDepartment of RadiologyMedical University of ViennaVienna, Austria

Stuart E. Mirvis, MDDepartment of RadiologyMaryland Shock-Trauma CenterUniversity of Maryland School

of MedicineBaltimore, Md

Fabian Moselewski, BSDepartment of RadiologyMassachusetts General HospitalBoston, Mass

Robert A. Novelline, MDDepartment of RadiologyMassachusetts General HospitalBoston, Mass

Marcel O. Philipp, MDDepartment of RadiologyMedical University of ViennaVienna, Austria

James T. Rhea, MDDepartment of RadiologyMassachusetts General HospitalBoston, Mass

KathirkamanathanShanmuganathan, MD

Department of RadiologyUniversity of Maryland School of

MedicineBaltimore, Md

Carlos J. Sivit, MDCase Western Reserve University

School of MedicineRainbow Babies and Children’s

HospitalCleveland, Ohio

D. Christian Sonne, MDDepartment of RadiologyUniversity of California, San FranciscoSan Francisco, Calif

A. Gregory Sorensen, MDDepartment of RadiologyMassachusetts General HospitalBoston, Mass

Eric J. Stern, MDDepartment of RadiologyHarborview Medical CenterUniversity of WashingtonSeattle, Wash

Nisa Thoongsuwan, MDDepartment of RadiologyHarborview Medical CenterUniversity of WashingtonSeattle, Wash

Lacey Washington, MDDepartment of RadiologyMedical College of WisconsinMilwaukee, Wis

O. Clark West, MDDepartment of RadiologyUniversity of Texas Medical School

at HoustonHouston, Tex

5

Preface

Why a categorical course on Emergency Radiology? Although it has been a while since the

last such course, the curriculum responds to the manifold challenges of community-

focused health care organization that have been caused by accelerating demographic,

cultural, and technologic changes. Interactions between structural and fiscal changes in

health care delivery and financing have altered the traditional triage and clearing-house

roles of emergency departments, which increasingly provide primary care and definitive

diagnoses and treatments. The resultant overburdening of emergency departments leads

to pressures to shorten the length of the in-facility stay, which begets greater use of

adjunct tests, and pressures to obtain and interpret images quickly—not to mention

accurately—24/7. Enter enabling imaging technologies (eg, multi–detector row CT)

and information technologies (eg, PACS, RIS).

Substantial synthesis of new and well-found knowledge is reviewed in this syllabus,

and even more will be presented during the course by an exceptional faculty. None of

this could or would occur without the leadership and support of the RSNA. In particular,

I especially need to acknowledge the assistance and forbearance of Robert A. Novelline,

MD, Chair of the RSNA Refresher Course Committee, and Mss Diane Lang, Ann Blair,

Annette Savage, and Eileen Brazelton.

I can only hope you learn and enjoy reading these chapters as much as I have. Best

wishes.

Frederick A. Mann, MD

6

7

The Emergence ofEmergency Radiology1

Emergency radiology is the subspecialty of radiology that deals with the imaging ofacutely ill and injured patients and the imaging management of such cases (1). Emer-gency radiology is one of the newest subspecialties of radiology and one of the fastestgrowing. In recent years, emergency radiology has emerged into the spotlight. The cur-rent drive to improve quality of care and decrease health care costs is demanding fasterand more sophisticated diagnostic imaging of patients at emergency centers at the sametime that the volume of patients seeking care at emergency centers is increasing. In addi-tion, the demand for immediate off-hours radiologic interpretation for emergency casesis changing the practice of radiology.

According to the National Center for Health Statistics of the U.S. Department ofHealth and Human Services (2), currently more than 100 million U.S. residents (110.2million in 2002) come annually to an emergency center for evaluation of an acute con-dition. The annual number of U.S. injury-related visits is currently more than 39 million(39.2 million in 2002). Nearly all patients seen in an emergency department undergo atleast one imaging examination as part of their diagnostic evaluation, and many willundergo several, including radiographic, ultrasonographic (US), computed tomographic(CT), and magnetic resonance (MR) imaging examinations. Emergency imaging exami-nations must be of the highest quality, and images must be obtained and interpreted ina timely manner so that a quick and accurate diagnosis can be made.

The National Center of Health Statistics (2) has reported a steady increase in the vol-ume of patients seen at emergency departments. For 1996, the center reported 34 annualvisits per 100 persons; in 2002, this number had increased to 39 visits per 100 persons.In the period from 1995 to 1996, there were 36 million injury-related emergency depart-ment visits recorded in the United States, and today that annual number is more than39 million.

At Massachusetts General Hospital, we have been observing a steady 5% annual in-crease in the volume of patients seen at the emergency department during the past de-cade. Currently, our emergency department treats more than 80.000 patients annually,and emergency radiology personnel perform nearly 80.000 imaging examinations peryear, approximately one imaging examination per patient.

To provide emergency imaging services for this patient population, the EmergencyRadiology Division of Massachusetts General Hospital is equipped with three digitalradiography rooms, two multi–detector row CT scanners, a 1.5-T MR imager, and USequipment. Imaging in the division is completely digital, with five picture archivingand communication system (PACS) workstations and a three-dimensional workstationfor image management and consultations. In-house staff radiologists provide interpreta-tions 24 hours per day, 7 days per week. The Emergency Radiology Division also serves

Robert A. Novelline, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 7–9.

1From the Department of Radiology, Harvard Medical School, Massachusetts General Hospital, 32 Fruit St, PO Box9657, Boston, MA 02114 (e-mail: rnovelline@partners.org).

No

velli

ne

8

the radiologic needs of the remainder of the medicalcenter during off hours, performing a wide array ofemergency inpatient and outpatient imaging examina-tions during evenings and weekends.

Fortunately, there have been spectacular advances indiagnostic imaging during the past decade to meet thedemands placed on emergency radiology. CT (3) alonehas revolutionized imaging diagnosis for trauma pa-tients, and a new revolution is currently under waywith multi–detector row CT scanners, which can scanpatients faster, with thinner sections, at higher spatialresolution, and with less radiation exposure than ear-lier scanners.

Ten years ago, it took 20 minutes to perform a CTscan of the head; today a 16–detector row CT examina-tion of the head can be performed in less than 10 sec-onds. Ten years ago, imaging a patient with multipletrauma who needed CT scans of the head, cervicalspine, chest, abdomen, and pelvis would have required1–2 hours in the CT scanner suite, including the frus-trating 5–10-minute waiting periods between scansegments to allow the x-ray tube to cool. This length oftime in the radiology area was potentially hazardousfor trauma patients and required valuable nursing andphysician staff-hours to manage trauma cases in theradiology area.

Today, a “total-body” trauma scan can be performedin less than 5 minutes (4). The 16–detector row CTscanners are so fast that one actually has to slow downthe scanning of trauma patients with programmed de-lays so that the scanner will not scan ahead of themovement of the intravenously administered contrastmaterial. In the past, many trauma patients who wereseriously injured or were in potentially unstable condi-tion were not able to spend long periods of time in theradiology area to receive the benefits of CT diagnosis;most of these patients today can be examined withmulti–detector row CT. Because trauma is the leadingcause of death of U.S. citizens younger than 40 yearsold and is the third most common cause of death ofall U.S. citizens (5), the advances in CT diagnosis fortrauma cases can profoundly affect U.S. health care.

In addition to the increased speed, the advances inmulti–detector row CT with 16–detector row CT scan-ners have also provided dramatic improvements in CTreformations (two-dimensional, three-dimensional,and curved planar reformations), CT angiography, andCT perfusion and cardiac-gated CT examinations. Sagit-tal and coronal reformations of the spine of trauma pa-tients examined with chest-abdomen CT are so high inquality that thoracolumbar spine radiographs are nolonger required. Coronal and sagittal multi–detectorrow CT reformations now are used to assist in the diag-nosis of facial fractures, sternal fractures, diaphragmrupture, gallbladder avulsion, pelvic fractures, anddozens of other injuries not optimally depicted in thetransverse plane. Three-dimensional reformations of

fractures provide excellent displays to show the posi-tion and alignment of fracture fragments.

With 16–detector row CT technology, CT angiogra-phy is so improved that it is replacing lengthy conven-tional arteriography as the imaging modality of choicefor patients suspected of having injuries of the aortaand major blood vessels of the head, neck, chest, abdo-men, pelvis, and extremities. Cardiac gating of chest CTperformed because of trauma can eliminate the aorticpulsatile motion artifact that has been responsible formany indeterminate interpretations of CT images ob-tained in patients suspected of having aortic trauma.Perfusion CT imaging can be used to diagnose cerebralischemia associated with traumatic dissection or otherinjuries of the carotid or vertebral arteries and has po-tential for use in diagnosing traumatic ischemia ofother organs, such as the kidneys.

CT has also revolutionized the work-up of non-traumatic emergency conditions and currently is thediagnostic procedure of choice in most emergencycenters for patients suspected of having appendicitis,diverticulitis, renal stone disease, bowel obstruction,aortic dissection, aortic aneurysm, or pulmonary em-bolism (3). CT performed in the emergency center canbe used (a) to expedite management and treatment byproviding confirmation when disease is present and(b) to prevent unnecessary surgery and/or hospitaliza-tion by identifying patients without disease or thosewith an alternative diagnosis. The 16–detector row CTscanners can scan faster than four–detector row orsingle–detector row CT scanners do and produce ex-cellent scans with less patient radiation, a markedbenefit for children and pregnant woman who requirean emergency CT diagnosis.

Advances in MR imaging also have been of great di-agnostic value in the emergency department. MR is theimaging modality of choice for the diagnosis of acuteischemic stroke and contributes substantially to the di-agnosis of other acute conditions of the central nervoussystem. MR imaging is indicated for patients withspinal trauma who have neurologic deficits and forpatients with acute musculoskeletal trauma when theconventional radiographic findings are indeterminateor when a soft-tissue injury is suspected. MR is also in-dicated for the diagnostic imaging of acute aortic andother vascular conditions in patients who cannot be ex-amined with contrast material–enhanced CT because ofa history of allergic reaction to contrast material or im-paired renal function. Recently developed fast MR im-aging protocols may play a future role in the diagnosisof acute thoracic and abdominal conditions in childrenand pregnant woman as an alternative to examinationsthat use ionizing radiation. Today, MR imagers aremore readily available to patients in the emergencydepartment.

Emergency US is another valuable technique in themanagement of cases in the emergency department.

Emerg

ence o

f Emerg

ency Rad

iolo

gy

9

US is the procedure of choice for patients suspected ofhaving deep venous thrombosis, acute cholecystitis,painless jaundice, or hydronephrosis. US is indicatedfor nearly all acute gynecologic, obstetric, and acutescrotal conditions. A distinct benefit of US in thetrauma patient who is in unstable condition is the factthat US can be performed at the bedside to screen forintraperitoneal hemorrhage and injuries to the majorabdominal organs.

Fortunately, digital imaging and PACS have devel-oped in concert with recent advances in cross-sectionalimaging. A typical total-body trauma scan of the head,cervical spine, chest, abdomen, and pelvis with coronal,sagittal, and other reformations may produce from 500to 700 images, and these unusually large data sets re-quire viewing each section with soft-tissue, bone, andlung windows. It would be incredibly burdensome toprint hard-copy film images of all of these images atvarious windows and then view the film images on lightboxes. Panning through such large image series onPACS while continuously changing windows acceleratesthe viewing and interpretation of images from traumaand other emergency cross-sectional examinations.

In many medical centers, a considerable number ofhospital admissions are arranged today through theemergency department, with patients having their initialimaging work-up done and the imaging diagnosis madein the emergency radiology section. Consequently,emergency radiology is becoming one of the important“diagnostic centers” of the institution. At MassachusettsGeneral Hospital, 46% of the inpatients are admittedthrough the emergency department, with their initialdiagnostic imaging examinations performed and inter-preted in the emergency radiology division.

Off-hours coverage of emergency radiology servicesis provided today either by an on-site radiologist orthrough teleradiology. On-site coverage is easier to ar-range in academic centers, where the off hours are usu-ally covered by evening, night, and weekend residents.The residents provide preliminary off-hours interpreta-tions that are subsequently checked by staff on the nextmorning. However, corrections made during “over-reads” of resident interpretations may lead to delays incorrect diagnosis and delays in patient treatment. Con-sequently, many academic centers are now institutingfull-time in-house staff coverage by using “nighthawk”staff radiologists (staff radiologists working the nightshift). Off-hours coverage at nonacademic centers ismore difficult to provide because of the current short-age of radiologists. Some private groups do now pro-vide 24-hour staff coverage or have contracted withteleradiology nighthawk services.

The demands for radiologists with specialization inemergency radiology and the professional opportuni-ties available today have stimulated great interest inthis subspecialty as a career choice. The American So-ciety of Emergency Radiology (ASER) was founded in

1988 to serve the needs of those interested in the fieldof emergency radiology. ASER now has more than 450U.S. and international members and an annual scien-tific meeting with more than 200 registrants. ASERpublishes the journal Emergency Radiology: A Journal ofPractical Imaging and has sponsored a core curriculumin emergency radiology for residents, medical students,and fellows in emergency radiology. The specialty ofemergency radiology has been recognized by theAmerican College of Radiology, which has offered aseat on the council for an ASER representative. Boththe Radiological Society of North America and theAmerican Roentgen Ray Society have recognizedemergency radiology in the planning of scientific ses-sions and instructional refresher courses in emergencyradiology, as well as including sections of emergencyradiology in their respective journals, Radiology andAJR: American Journal of Roentgenology.

Research in emergency radiology has involvedboth retrospective and prospective investigations ofnew imaging modalities and protocols for the diag-nosis of trauma and nontraumatic emergency condi-tions. Because of the current concerns about risinghealth care costs and possible overuse of imaging re-sources, a recent direction in research is the investiga-tion of (a) criteria for selecting patients for emergencyimaging examinations, such as the National EmergencyX-Radiography Utilization Study (6); or (b) criteria (7)to select the most appropriate imaging examinationonce the decision to image has been made.

As chairman of the RSNA Refresher Course Com-mittee, it gives me great pleasure to invite you to at-tend the RSNA 2004 Categorical Course in DiagnosticRadiology: Emergency Radiology and/or read this cat-egorical course syllabus prepared by the participants.This course will cover the spectrum of emergency radi-ology, highlighting cutting-edge technologies and im-aging approaches. I am most grateful to all of the in-vited speakers, who have prepared excellent writtenmaterials and presentations for this course.

References1. Harris JH Jr. Reflections: emergency radiology. Radiology 2001;

218:309–316.2. Emergency department visits. NCHS FASTATS page. National

Center for Health Statistics Web site. Available at: www.cdc.gov/nchs/fastats/ervisits.htm. Accessed April 16, 2004.

3. Novelline RA, Rhea JT, Rao PM, Stuk JL. Helical CT in emergencyradiology. Radiology 1999; 213:321–339.

4. Ptak T, Rhea JT, Novelline RA. Experience with a continuous,single-pass whole-body multidetector CT protocol for trauma: thethree minute multiple trauma CT scan. Emerg Radiol 2001; 8:250–255.

5. National Safety Council. Injury facts: 2003 edition. Itasca, Ill:National Safety Council, 2003.

6. Hoffman JR, Wolfson AB, Todd K, Mower WR. Selective cervicalspine radiography in blunt trauma: methodology of the NationalEmergency X-Radiography Utilization Study (NEXUS). Ann EmergMed 1998; 32:461–469.

7. Blackmore CC, Emerson SS, Mann FA, Koepsell TD. Cervicalspine imaging in patients with trauma: determination of fracture riskto optimize use. Radiology 1999; 211:759–765.

NO

TES

11

Nontraumatic NeurologicEmergencies1

There are a few life-threatening neurologic illnesses in which emergent imaging canplay a key role. This chapter will describe the role of imaging in these diseases and alsowill highlight some of the common neurologic illnesses that are identified with neuro-imaging in the acute-care setting of the modern North American emergency depart-ment.

Unfortunately, the most common diseases of the brain are not yet amenable to as-sistance with neuroimaging. These diseases include the neuropsychiatric illnesses, suchas major depression, bipolar disorder, and schizophrenia, and alcoholism, substanceabuse, and addiction. It may be that in the future, our imaging tools will provide uswith insight into these diseases and will be used to assist in their acute management.Another class of acute neurologic illness, trauma, is covered in other chapters of thissyllabus. This leaves still a sizable number of neurologic illnesses; this syllabus chapterwill focus only on the most common. These include stroke (ischemic and hemor-rhagic), stroke mimics (such as posterior leukoencephalopathy syndrome), acutedemyelinating disease, new adult-onset seizures, and infection. Finally, because acommon request for imaging is to determine the safety of proceeding with lumbarpuncture to sample cerebrospinal fluid, a short discussion of this topic will also bepresented.

STROKE

Stroke is a term that describes a rapid onset of neurologic impairment. Eighty-five per-cent of all strokes are ischemic stroke, and the vast majority of these are caused byblockages of arterial flow to brain tissue, with subsequent cellular impairment and/ordeath. Fifteen percent of all strokes are hemorrhagic; this includes both intraparenchy-mal hemorrhage and subarachnoid hemorrhage, with the latter most commonlycaused by rupture of an intracranial aneurysm. For each of these entities, the patientdeserves emergent neuroimaging for diagnosis and treatment planning.

A. Gregory Sorensen, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 11–15.

1From the Department of Radiology, Massachusetts General Hospital, A. A. Martinos Center, Bldg 149, 13th St, Charles-town, MA 02129 (e-mail: sorensen@nmr.mgh.harvard.edu).

A.G.S. receives research support from, is a consultant for, or has spoken on behalf of the following companies within thelast year: Siemens Medical Systems; General Electric Medical Systems; Glaxo SmithKline; Novartis Pharmaceuticals;Descartes Therapeutics; Schering AG; Hemedex, Inc; Pfizer, Inc; StemCells, Inc; and Transkaryotic Therapies, Inc.

In addition, A.G.S. has an equity position in and holds the position of Medical Director at EPIX Medical, Inc, a specialtypharmaceutical company based in Cambridge, Mass, engaged in developing targeted contrast agents for cardiovascularMR imaging.

Sore

nse

n

12

Ischemic Stroke

Stroke occurs about 700.000 times per year in theUnited States and is the leading cause of adult morbid-ity and a leading cause of death. In the past 10 years,the treatment of stroke has begun to shift from initialpalliative care, followed by an emphasis on secondaryprevention, to a focus on acute treatment, with second-ary preventive measures still important but not emer-gent. This evolution in therapy is due to the approval in1996 by the Food and Drug Administration of a throm-bolytic agent, alteplase (recombinant tissue plasmino-gen activator, or rt-PA). The results of a number of well-controlled clinical trials have demonstrated that ad-ministration of alteplase leads to decreased morbidityand mortality when treatment is initiated within 3hours from the onset of symptoms (1) (Fig 1).

Imaging plays a crucial role in alteplase therapy fortwo reasons. First, alteplase must not be given if thestroke is hemorrhagic, rather than ischemic. Althoughthese two types of stroke can often be distinguishedclinically, imaging is still required to rule out hemor-rhage prior to administering alteplase. Because com-puted tomography (CT) is typically available rapidly inthe emergency setting and because it is widely acceptedas a very sensitive tool for the detection of acute hemor-rhage, CT is the standard initial imaging modality.Naturally, any sign of hemorrhage at CT is a contraindi-cation to thrombolytic therapy. This is because the rateof symptomatic hemorrhage increases in stroke patientstreated with alteplase from 1% to 7% or more. Hemor-rhage is a feared complication of alteplase administra-tion and therefore is the main sign sought at the initialCT examination. However, there are additional signs atCT that are considered to be contraindications for acutechemical thrombolysis. These signs are all ways of esti-mating the size and the age of the infarct.

Current thinking is that the risk of hemorrhage ishigher in the larger and more advanced infarcts, prob-ably because of the increased chance of reperfusion in-jury accompanied by symptomatic hemorrhage. For ex-ample, a standard rule of thumb is that treatment withalteplase is contraindicated if more than one-third of themiddle cerebral artery territory of the brain has evidenceof low attenuation on CT images obtained at the acuteCT examination. An example of this is shown in Figure2. Hemorrhage can be a complication of ischemic strokeeven without the administration of any thrombolytictherapy. Attention to the window level settings may pro-vide greater sensitivity to subtle signs of stroke (2). Othersigns of severe stroke, such as midline shift, are alsothought to represent contraindications to therapy.

Although no well-controlled trials have been per-formed to demonstrate the benefit of adding CT angi-ography to the initial imaging study, many practitio-ners do add CT angiography because it is thought toidentify clot, if present, in the proximal intracerebral ar-

teries. The presence of clot can confirm the cause of thestroke. However, it is not yet clear to what degree theinitial CT angiographic findings correlate with outcomeor to what extent they should be used to guide treat-ment because CT angiography does not always allowgood evaluation of the distal collateral vessels.

Although CT has become the accepted standard forimaging acute stroke, many groups are also exploringthe utility of magnetic resonance (MR) imaging inacute stroke. MR imaging is being explored for a num-ber of reasons. One is to distinguish infarction fromtransient ischemic attack. Transient ischemic attack isnow also considered a medical emergency because ofthe high frequency of cerebrovascular events aftertransient ischemic attack (3). Identifying the cause ofa transient ischemic attack, such as a dissection (Fig3), can assist in the work-up and prevention of thetransient ischemic attack culminating in a full-fledgedinfarct. Diffusion-weighted MR imaging is the mostsensitive and specific single imaging modality for di-agnosis of stroke, and the identification of tissue dam-age at neuroimaging, even if symptoms have quicklyresolved, will have therapeutic implications.

A second reason for the use of MR imaging is that,despite the confirmed efficacy of chemical thromboly-sis, only a small fraction (estimated at 5%) of strokevictims receive alteplase. This is in part because manypatients are outside the 3-hour window of time, afterwhich the risk of hemorrhage goes up substantially.Nevertheless, there have been suggestions that manypatients may have lesions that would still be ame-nable to treatment at this later time, and MR imagingis thought to be the way to identify such patients. Theuse of MR imaging, particularly identification of a dif-fusion-perfusion mismatch, is believed to suggest thepresence of salvageable tissue. Although this conceptis still somewhat the subject of debate, most practitio-ners believe that a large diffusion-perfusion mismatch

Figure 1. Model-estimated odds ratio (OR) for favorable out-come at 3 months in alteplase-treated patients, compared withcontrols, by time from onset of symptoms (OTT). Odds ratiowas adjusted for age, baseline glucose concentration, baselineNational Institutes of Health Stroke Scale (NIHSS) score, base-line diastolic blood pressure, previous hypertension, and inter-action between age and baseline NIHSS measurement. (Re-printed, with permission, from reference 1.)

No

ntrau

matic N

euro

log

ic Emerg

encies

13

suggests the need for more aggressive treatment. Theconcept of the ischemic penumbra is similar to the so-called stunned myocardium in the heart, and the devel-opment of imaging tools to study this has been a majoreffort during the past decade.

A third use for emergent MR imaging in stroke iswhen there is a question of venous thrombosis. MRvenography and CT venography have not been for-mally compared in any well-controlled trials that

studied patients with acute stroke (4,5), but mostpractitioners, extrapolating from other data, believethat MR venography may have higher accuracy.

Hemorrhagic Stroke

The 15% of all strokes that are due to hemorrhage canbe categorized as either parenchymal or subarachnoid.Both types are easily distinguished at CT in routine prac-tice, and a review of these findings is not necessary here.

Figure 3. Dissection of combined extra- and intracranial carotid artery. Left: Two-dimensional time-of-flight MR angiogram showsabsence of right internal carotid artery above bifurcation. Top center: Three-dimensional time-of-flight MR image of circle of Willisshows attenuation of flow in right middle cerebral artery and absent flow in right distal internal carotid artery. Bottom center: Two-dimensional phase-contrast MR angiogram of circle of Willis shows cross-filling and retrograde flow through precommunicating partof right anterior cerebral artery (A1 segment). Right: T1-weighted MR image with fat saturation shows hyperintense thrombus in ves-sel wall in right internal carotid artery in neck just below skull base.

Figure 2. Left: CT images from initial CT examination of the head performed 2 hours 45 minutes after the onset of symptoms in a 42-year-old man show hypoattenuation covering more than one-third of the middle cerebral artery territory. Right: CT images from follow-upCT scan performed 6 hours after onset of symptoms show hemorrhagic transformation with midline shift.

Sore

nse

n

14

There are a few important updates: First, CT angiogra-phy increasingly is being used in the emergency set-ting to evaluate the location and nature of the aneu-rysm. Familiarity with CT angiography and with thelocations of aneurysms is therefore increasingly be-ing driven into the emergency department, ratherthan the neuroangiography suite. A second importantshift is due to changes in clinical practice driven bythe results from the International Subarachnoid Aneu-rysm Trial. This study compared 1-year clinical out-comes after either coiling or clipping of aneurysms.The group undergoing clipping had a substantiallygreater morbidity at 1 year. Although this appears toshow clinical superiority for the less-invasive ap-proach, the clearest assessment of the benefits of coil-ing will come after the 5-year follow-up data are as-sembled. This further suggests that radiologists willplay an important role in the diagnosis and manage-ment of aneurysmal disease.

STROKE MIMICS

A number of disease processes can mimic the appear-ance of stroke, especially at CT. Fortunately, knowl-edge of the clinical manifestations of these illnesses,combined with a clear understanding of the mostpowerful tool in the diagnostic armamentarium foracute ischemic stroke, that is, diffusion-weighted MRimaging, will allow a clear diagnosis in most cases.

One relatively recently recognized syndrome is thattermed reversible posterior leukoencephalopathy syn-drome (6). Although initially thought to be a compli-cation of malignant hypertension and/or associatedwith toxemia of pregnancy, this syndrome has nowbeen described in subjects with a range of illnesses. Be-yond hypertension and toxemia, these additional ill-nesses include chronic renal disease, pheochromocy-

toma, and acute glomerulonephritis, and the syndromehas been seen as a side effect of chemotherapy. Clinicalreversible posterior leukoencephalopathy syndrome istypically accompanied by headache or other symptomsranging from nausea and/or vomiting to convulsions,stupor, or coma. On pathologic examination, micro-scopic hemorrhages and infarcts may be evident, al-though early MR imaging shows increases in the appar-ent diffusion coefficient, rather than reductions. Lower-ing the blood pressure is the major form of treatmentand should be done before stroke (hemorrhagic or is-chemic) occurs. There have been some documentedcases of reversible posterior leukoencephalopathy syn-drome without hypertension.

There is a range of other stroke mimics: lesions thatmay have similar findings with one or more types ofMR imaging. However, careful analysis of imaging,particularly diffusion-weighted images, usually can beused to sort these out. One particularly important stepis understanding the difference on diffusion-weightedimages between hyperintense signal intensity causedby restricted water mobility—that is, lowered valuesfor the apparent diffusion coefficient—and hyperin-tense signal intensity on diffusion-weighted imagesdespite normal or even elevated water mobility. Thiscan occur when a region has such hyperintense signalintensity on T2-weighted MR images that even afterthe diffusion encoding gradients are applied, there isstill residual hyperintense signal intensity: the so-called T2 shine-through effect.

OTHER ACUTE NEUROLOGIC ILLNESSES

Although stroke remains the most common neurologicemergency, a variety of other illnesses can be diagnosedfor the first time in the emergency department with theaid of imaging. One of these is acute demyelinating

Figure 4. Brain abscess in 33-year-old woman with 3-day history of vomiting and seizure. A, T1-weighted MR image obtained af-ter administration of gadolinium-based contrast material shows ring enhancement. B, T2-weighted MR image shows edema sur-rounding mass, and differential diagnosis includes both abscess and tumor. C, Diffusion-weighted MR image shows hyperintensityin the center of the abnormality. D, Map of apparent diffusion coefficient shows low apparent diffusion coefficient in center of abnor-mality. Surgical drainage demonstrated yellow pus that grew streptococci.

No

ntrau

matic N

euro

log

ic Emerg

encies

15

disease. Demyelinating disease is located typically inthe white matter and can be due to either multiplesclerosis or acute disseminated encephalomyelitis,and these entities can look identical. Although the for-mal diagnosis of multiple sclerosis requires “multiple”events (ie, dissemination of lesions in time and space[7]), many practitioners do not always wait for themultiple events to occur before beginning presump-tive treatment for multiple sclerosis with the new dis-ease-modifying agents, such as the interferon-betaclass. Although optic neuritis can be an isolated find-ing, the frequency of this progressing to multiple scle-rosis is high enough to lead to immediate treatment,with tapering of the treatment if no symptoms occur.Acute disseminated encephalomyelitis is thereforeusually distinguished by history: There is usually ahistory of immunization and/or a viral prodrome (8).

Intracranial abscesses or encephalitis can also occur(9), but these findings are relatively uncommon(about 1 in 100.000 per year) (10). In modern series,intracranial abscesses and encephalitis are typicallyfound in immunocompromised patients, and theusual infections are from toxoplasmosis, Nocardia, etc.Typical causes include contiguous spread (eg, from se-vere otitis media), hematogenous spread, or a sequelaafter some breach of physical defenses (eg, aftertrauma or neurosurgery). Some 20%–30% of these in-fections are cryptogenic. Diffusion-weighted MR imag-ing has been reported to show restricted water mobil-ity in the center of the abscess (Fig 4).

New onset of seizures is another neurologic illness inwhich imaging can play a role in the emergency depart-ment. However, in this case, the role is to exclude ana-tomic lesions amenable to focused intervention be-cause most new seizures are idiopathic (some 62% inone series) (11). However, vascular disease (eg, stroke)is frequent in older patients (49% of the older patientswith new seizures in that same series), and tumor ac-counted for 11%. Therefore, routine MR imaging canbe helpful in both ruling in and ruling out disease.

CAN I PERFORM LUMBAR PUNCTURE?

Although neuroimaging has become more commonthan lumbar puncture for neurologic diagnosis, cere-brospinal fluid sampling is still required for manywork-ups. If the intracranial pressure is raised or if amass is blocking the normal flow of cerebrospinal

fluid, lumbar puncture can result in herniation (12).As a result, some clinicians seek to rule out suchmasses in clinical settings in which herniation isthought to be a risk. Our role in such settings is to ex-clude a compartment syndrome. Because the totaldaily volume of cerebrospinal fluid production isabout three times that of the total cerebrospinal fluidvolume (total volume in an adult is about 150 mL,with daily production of about 450 mL), if the flow ofcerebrospinal fluid from one compartment to anotheris blocked, a compartment syndrome can arise. Hence,to clear a patient for lumbar puncture simply meansto ensure that no blockages to flow are present; this ismost easily ascertained by a lack of brain shift.

References1. Hacke W, Donnan G, Fieschi C, et al. Association of out-

come with early stroke treatment: pooled analysis ofATLANTIS, ECASS, and NINDS rt-PA stroke trials. Lancet2004; 363:768–774.

2. Lev MH, Farkas J, Gemmete JJ, et al. Acute stroke: im-proved nonenhanced CT detection—benefits of soft-copyinterpretation by using variable window width and centerlevel settings. Radiology 1999; 213:150–155.

3. Albers GW, Caplan LR, Easton JD, et al. Transient is-chemic attack: proposal for a new definition. N Engl J Med2002; 347:1713–1716.

4. Ciccone A, Canhao P, Falcao F, Ferro JM, Sterzi R. Throm-bolysis for cerebral vein and dural sinus thrombosis.Cochrane Database Syst Rev 2004; 1:CD003693.

5. Campbell BG, Zimmerman RD. Emergency magnetic reso-nance of the brain. Top Magn Reson Imaging 1998; 9:208–227.

6. Mukherjee P, McKinstry RC. Reversible posterior leukoen-cephalopathy syndrome: evaluation with diffusion-tensorMR imaging. Radiology 2001; 219:756–765.

7. Poser CM, Brinar VV. Diagnostic criteria for multiple sclero-sis: an historical review. Clin Neurol Neurosurg 2004; 106:147–158.

8. Gabis LV, Panasci DJ, Andriola MR, Huang W. Acute dis-seminated encephalomyelitis: an MRI/MRS longitudinalstudy. Pediatr Neurol 2004; 30:324–329.

9. Tattevin P, Bruneel F, Clair B, et al. Bacterial brain ab-scesses: a retrospective study of 94 patients admitted to anintensive care unit (1980 to 1999). Am J Med 2003; 115:143–146.

10. Das P. Infectious disease surveillance update. Lancet InfectDis 2004; 4:259.

11. Sander JW, Hart YM, Johnson AL, Shorvon SD. NationalGeneral Practice Study of Epilepsy: newly diagnosed epi-leptic seizures in a general population. Lancet 1990; 336:1267–1271.

12. van Crevel H, Hijdra A, de Gans J. Lumbar puncture andthe risk of herniation: when should we first perform CT?J Neurol 2002; 249:129–137.

NO

TES

17

Traumatic Brain Injury:Imaging Update 20041

Trauma continues to be the number one cause of death in individuals younger than44 years old (1). In the United States alone, the cost of head trauma to the public isestimated to be $40 billion annually (2). During the past decade, considerable ad-vances have been made in the field of neuroimaging of these patients. For computedtomography (CT), the acquisition time has decreased with the development of heli-cal, and now multi–detector row, CT scanners. Magnetic resonance (MR) imaginghas become more accessible, with faster acquisition times for pulse sequences, newersequences with greater sensitivity for subtle abnormalities, and the potential forevaluating functional abnormalities subsequent to trauma.

Indeed, apart from imaging acute subarachnoid hemorrhage and skull base frac-tures, MR imaging is superior to CT in the depiction of virtually all manifestations oftraumatic brain injury. The lack of beam-hardening artifacts, combined with theability to perform multiplanar reconstructions, allows MR detection of small extra-axial collections, which is particularly useful in identifying the imaging manifesta-tions of child abuse or domestic violence (3). MR imaging can also be used to tem-porally stage intracranial hemorrhage, which may be important in diagnosing abusebecause multiple sites of hemorrhage at different stages of evolution suggest recur-rent trauma. In addition, the signal abnormality caused by prior hemorrhage is seenon MR images far longer than the attenuation abnormality seen on CT images, thusallowing detection of lesions caused by earlier abuse. Further, white matter shearinginjuries often are identified only with MR imaging. The lack of ionizing radiationwith MR imaging is an added bonus, particularly if serial studies are necessary.

More recently, MR spectroscopy, MR diffusion imaging, and functional imagingtechniques such as positron emission tomography (PET), single photon emission CT(SPECT), and functional MR imaging have yielded insights into traumatic brain in-jury. These modalities are likely to have a prognostic role in the care of trauma pa-tients. In spite of these MR imaging advantages, CT continues to be the initial studyof choice for evaluating acute traumatic brain injury.

In this chapter, we review the current imaging approach and imaging findings inadult traumatic brain injury. We also discuss the newer techniques that may not nec-essarily alter the acute management of traumatic brain injury but are likely to leadus to a better understanding of its pathophysiology.

CLASSIFICATION OF TRAUMATIC BRAIN INJURY

There are several ways to classify traumatic brain injury (4,5). First, one can divideacute head injuries into primary and secondary lesions. A primary lesion occurs at the

Alisa D. Gean, MD, Christine Glastonbury, MBBS,and D. Christian Sonne, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 17–32.

1From the Department of Radiology, University of California, San Francisco, 1001 Potrero Ave, San Francisco, CA 94110(e-mail: alisa.gean@radiology.ucsf.edu).

Gea

n e

t al

18

time of injury, as a direct result of the traumatic force(eg, cortical contusion, skull fracture, cranial nerve in-jury, white matter shearing injury). A secondary lesiongenerally occurs as a consequence of a primary lesion(eg, vasospasm, hypoxia, seizures, and intracranial hy-pertension from expanding mass lesions or edema).This division is clinically important because second-ary injury is often preventable, whereas primary le-sions have already occurred by the time the patient isfirst seen in the emergency department. This classifica-tion also underscores the fact that traumatic brain in-jury is not a static event but is, in fact, a progressive in-jury that is complex at the cellular level.

Traumatic brain injury may also be classified ac-cording to lesion location (ie, intra- vs extraaxial),mechanism of injury (penetrating/open vs blunt/closed), and clinical severity (minor, mild, moderate,or severe). Eighty percent of the injuries classified astraumatic brain injury consist of mild head injuries(defined as a score on the Glasgow coma scale ≥ 13),10% are moderate (Glasgow coma score, 9–12), andanother 10% are severe (Glasgow coma score, 3–8).

WHO SHOULD BE IMAGED?

Emergency CT of the head in the setting of acute traumais indicated for the following criteria: (a) Glasgowcoma score less than 8 or a decrease of more than 3 inthe Glasgow coma score, (b) persistent neurologicdeficit, (c) anterograde amnesia, (d) unexplained pu-pillary inequality, (e) prolonged loss of consciousnessfor longer than 5 minutes, (f) depressed skull fracture,(g) penetrating injury, or (h) a bleeding diathesis oranticoagulant therapy (6). In the setting of minorhead injury (defined as Glasgow coma score = 15,loss of consciousness, and normal findings on briefneurologic examination), less than 10% of the pa-tients have positive CT findings, and less than 1% re-quire neurosurgical intervention (7,8). Investigatorsin one study suggested that in this group of patients,the presence of one or more of the following sevenclinical criteria requires CT scanning: (a) headache,(b) vomiting, (c) age older than 60 years, (d) drug oralcohol intoxication, (e) short-term memory deficits,(f) seizure, and (g) physical evidence of trauma abovethe clavicles (9).

HOW SHOULD THEY BE IMAGED?

Acute Traumatic Brain Injury

For the acutely injured patient, CT is the initial im-aging modality of choice. Its widespread accessibility,speed, relatively low cost, compatibility with life-sup-port devices, and CT angiographic capability make CTparticularly attractive for the evaluation of the criti-cally ill patient. Nonenhanced axial scans rapidly pro-vide accurate localization of space-occupying hemato-

mas and demonstrate the extent of mass effect, cister-nal compression, and/or hydrocephalus. CT allowsoptimal evaluation of the calvaria, skull base, and fa-cial bones within the same scan or with the additionof thin sections through a defined region. The intro-duction of the multi–detector row CT scanner allowsmore rapid evaluation of the head, neck, and body ofinjured patients, with markedly reduced imagingtimes and reduced scanner heat loading.

At San Francisco General Hospital, the routine CTprotocol for evaluating acute craniocerebral trauma isperformed without intravenous contrast material, andimages are viewed at three window settings: bone(level, 500 HU; width, 2000 HU), brain (level, 40 HU;width, 80–100 HU), and blood (level, 75 HU; width,150 HU). The digital lateral scout view, which serves asa “pseudo-skull film,” should always be assessed for askull fracture or upper cervical spine fracture that maynot be visible on the axial images. In rare situations,MR imaging may be recommended in the acute settingwhen neurologic findings remain unexplained after CT.

Subacute or Chronic Traumatic Brain Injury

Once the condition of the patient has stabilized, MRimaging may be indicated because of its superior sensi-tivity to the manifestations of traumatic brain injury.However, the widespread use of MR imaging in theacute setting is limited by the time involved in perform-ing MR imaging coupled with a reduced ability tomonitor a critically ill patient or one whose conditionis potentially unstable. MR imaging is more useful inthe subacute or chronic setting, in which its inherentincreased sensitivity to cortical contusions and shear in-juries is an important advantage over CT, particularly inthe evaluation of a patient whose condition is clinicallyworse than would be predicted from the CT findings.On routine T2-weighted MR images, inflammatorycells, edema, secondary ischemia, and certain stages ofblood products typically are hyperintense. Specific MRpulse sequences have been found to be more sensitivefor the depiction of these injuries and are described inmore detail subsequently. Functional MR imaging andnuclear medicine techniques such as PET and SPECTmay also be helpful in the nonacute setting becausethey offer insights into the pathophysiologic functionof trauma and the resultant functional deficits.

IMAGING CHOICES

Fluid-attenuated Inversion-Recovery Imaging

The fluid-attenuated inversion-recovery (FLAIR)pulse sequence increases the conspicuity of focal areasof increased T2 signal abnormality by eliminating (or“nulling”) the high signal intensity of cerebrospinalfluid. Thus, focal bright abnormalities of the gray mat-ter (eg, contusions) or white matter (eg, shear inju-ries) are more easily appreciated against adjacent dark

Traum

atic Brain In

jury

19

cerebrospinal fluid spaces (Fig 1). Sagittal and coronalFLAIR images are particularly helpful in the apprecia-tion of diffuse axonal injury involving the fornix andcorpus callosum, two areas that are difficult to distin-guish from adjacent cerebrospinal fluid on routine T2-weighted images (10). The FLAIR sequence also hasincreased sensitivity for the presence of acute or sub-acute subarachnoid hemorrhage, which appears asareas of hyperintensity within the sulci and cisterns(11). An important pitfall with FLAIR imaging is para-doxical cerebrospinal fluid hyperintensity in the sulciand cisterns of ventilated patients who are receiving ahigh inspired oxygen fraction of more than 0.60 (12).In addition, cerebrospinal fluid flow artifacts in thebasilar cisterns seen with FLAIR imaging can makeevaluation of brainstem pathologic abnormalitiessomewhat difficult.

Gradient-recalled Echo Imaging

The presence of ferromagnetic blood-breakdownproducts results in minute alterations in the localmagnetic field of tissue. Gradient-recalled echo (GRE)sequences are exquisitely sensitive to this field alter-ation, showing a marked drop in signal intensity, andthus making tiny hemorrhagic lesions readily appar-ent (13). Chronic blood-degradation products such ashemosiderin and ferritin remain for months to yearsafter injury, which makes the GRE sequence a power-ful tool in the evaluation of remote injury long afterabnormalities on T2-weighted and FLAIR images haveresolved. Currently, the GRE sequence is the mostwidely used MR sequence to detect parenchymal hem-orrhage, although conventional MR imaging is stillrelatively insensitive to microscopic changes of diffuseaxonal injury (14). Unfortunately, because of suscepti-bility artifact adjacent to the paranasal sinuses andmastoid air cells, GRE images are somewhat limited inthe evaluation of cortical contusions involving the in-ferior frontotemporal lobes.

Susceptibility-weighted Imaging

Several years ago, E. Mark Haacke, PhD, designed ahigh-spatial-resolution three-dimensional fast low-angle shot MR imaging technique that is extremelysensitive to susceptibility changes and, therefore, hem-orrhage (15). This sequence has subsequently beentermed susceptibility-weighted imaging. Researchersat Loma Linda University Medical Center have de-tected more shearing lesions with susceptibility-weighted imaging compared with conventional GREimaging (16). Although susceptibility-weighted imag-ing is not yet widely available, it holds promise in thediagnosis of the extent of diffuse axonal injury, as wellas providing valuable prognostic information. Specifi-cally, preliminary findings have shown that the extentof hemorrhage is strongly correlated with the initialseverity of injury and long-term outcome.

Diffusion-weighted Imaging and Diffusion-TensorImaging

A diffusion-weighted imaging sequence is a rapidlyacquired MR sequence that detects alteration in thefree mobility (or diffusibility) of water moleculesthrough tissues (17). Reduction or restriction of nor-mal diffusibility appears as increased signal intensityon diffusion-weighted MR images. The degree of dif-fusibility (or apparent diffusion coefficient) can becalculated and also represented in an image. Whenfluid motion is restricted, there is a low coefficient,and the region appears dark relative to normal tis-sues on the apparent diffusion coefficient map andbright on the diffusion-weighted MR image. Aniso-tropic diffusion (ie, preferential flow) is observed innormal myelinated white matter where diffusion isgreatest parallel to the fibers and lowest perpendicu-lar to them.

Diffusion-weighted imaging in traumatic brain in-jury can be used to depict abnormalities even whenconventional MR images are normal (18,19). In one

Figure 1. Role of FLAIR imaging in trau-matic brain injury. (a) Axial FLAIR MR im-age at level of centrum semiovale demon-strates a focal area of increased signal inten-sity within the left frontal subcortical whitematter (white arrow), consistent with a shear-ing lesion. A focal T2 hyperintense lesion(black arrow) is seen involving the cortex ofthe right frontal lobe, consistent with a corti-cal contusion, and a linear focus of increasedsignal intensity (circle) is seen within a leftparietal sulcus, consistent with subarachnoidhemorrhage. (b) On the corresponding axialT2-weighted spin-echo MR image, the le-sions are much less conspicuous.

Gea

n e

t al

20

study comparing conventional MR imaging (T2-weighted, FLAIR, and GRE imaging) with diffusion-weighted imaging in traumatic brain injury, investiga-tors found that diffusion-weighted imaging identifiedmore shearing injuries but was less sensitive for hem-orrhagic lesions than GRE (20).

Diffusion-tensor imaging capitalizes on the prin-ciples of the apparent diffusion coefficient and diffu-sion-weighted imaging because water diffuses morefreely along intact white matter fibers than acrossthem (ie, white matter anisotropy). Directionally en-coded color maps and three-dimensional tractographyare performed to assess fiber integrity. In one study ofpatients with mild traumatic brain injury and normalCT images obtained acutely (within 24 hours of in-jury), investigators showed focal areas of reducedanisotropy, particularly in the corpus callosum andinternal capsule (21). These abnormalities were lessapparent at 1 month in the two patients who under-went repeat imaging. The role of diffusion-weightedimaging and diffusion-tensor imaging in the care ofpatients needs further research, but these techniquesare providing insight into the pathophysiologic func-tion of traumatic brain injury.

Proton MR Spectroscopy

Proton MR spectroscopy (1H MR spectroscopy) al-lows noninvasive in vivo assessment of brain tissuethrough the quantification of cerebral metabolites(22). The primary metabolites include N-acetylaspar-tate, creatine, choline, glutamate, and lactate.N-acetylaspartate is a cellular amino acid and amarker for neuronal and axonal integrity. It is quanti-fied from a small selected volume of tissue (voxel)and is usually expressed as a ratio with respect to cho-line or creatine. Creatine, which produces a compositesignal consisting predominantly of creatine and phos-phocreatine, is considered to be a measure of cellulardensity and is especially high in glial cells (eg, post-traumatic gliosis). Increased choline signal may bedue to myelin injury and accumulation of membranemyelin-degradation products. Tissue injury is seen as areduction in the N-acetylaspartate–creatine ratio andmay be seen in injured brain within 24 hours aftertraumatic brain injury (23).

As is the case with the other previously mentionedMR techniques, MR spectroscopy is a tool that can beused to identify abnormalities in the setting of a nor-mal conventional MR examination. Specifically, lac-tate, a by-product of anaerobic glycolysis, may be dif-fusely elevated in otherwise normal-appearing brainand has been correlated with a poor clinical outcome.In addition, MR spectroscopy can be used to depict anabnormality in otherwise normal-appearing whitematter in the subacute and chronic period (24–26). Acorrelation has been shown between the reduction ofN-acetylaspartate, the elevation of choline, and the se-

verity of injury as measured with the Glasgow comascale or length of posttraumatic amnesia (24).

3-T Imaging

The relatively new 3-T imaging machines lacksome of the advantages that have evolved for 1.5-Timaging, including larger bores and wider fields ofview. Currently, claustrophobia and body size maylimit how many patients can benefit from imaging ina 3-T machine. In addition, because 3-T images aremore prone to susceptibility artifacts, surface contu-sions may be missed. Nevertheless, the increased sig-nal-to-noise ratio inherent in a 3-T machine isroughly double that at 1.5 T, and this higher signal-to-noise ratio can be used to reduce image acquisi-tion time and/or improve resolution. Advanced ap-plications, such as functional studies that are basedon blood oxygen level–dependent contrast, MR spec-troscopy, and diffusion-tensor imaging, are likely tobenefit from 3 T. Further, new phased-array coil sys-tems combined with “parallel imaging” techniqueoffer the promise of even faster imaging, which isparticularly helpful in the world of trauma imaging.

SPECT Imaging

Brain SPECT imaging is performed with a gammacamera 2 hours following the intravenous injection oftechnetium 99m hexamethylpropyleneamine oxime.The normal adult brain shows a bilateral symmetrictracer distribution, with higher activities in (a) tempo-ral, parietal, and occipital cortices, (b) basal nuclei,(c) thalami, and (d) the cingulate gyrus. Focal or re-gional areas of hypoperfusion may be evident in trau-matic brain injury and have been found to correlatebetter with the acute clinical status than with structuralimages (27). In another study, investigators found anassociation of acute areas of hypoperfusion with brainatrophy at 6 months, suggesting secondary ischemicdamage (28). Normal findings on brain SPECT studyhave been found to be reliable in the exclusion of theclinical sequelae of mild head injury.

PET Imaging

Much of the work in brain PET imaging has cen-tered on the cerebral utilization of glucose (as fluorine18 fluorodeoxyglucose). There is an increase in glu-cose metabolism following severe traumatic brain in-jury, which reflects the injury-induced energy crisis,with an increase in glucose utilization, a decrease inoxidative metabolism, and an uncoupling of cerebralblood flow. Some investigators have suggested agreater sensitivity of PET for the detection of abnor-malities in patients with mild and moderate traumaticbrain injury with normal MR studies who have persis-tent cognitive or behavioral complaints (29). Becausea nearby cyclotron is needed to generate the radioac-tivity, PET imaging is not widely available.

Traum

atic Brain In

jury

21

Functional MR Imaging

In functional MR imaging, the activity of the brainis observed through the detection of an alteration inthe ratio of cerebral blood deoxyhemoglobin to oxy-hemoglobin in response to particular tasks (30). Neu-ronal activation within the cerebral cortex leads to anovercompensation of blood flow relative to the in-crease in oxygen consumption, which results in a de-crease in capillary and venous deoxyhemoglobin con-centrations. Because deoxyhemoglobin is an endog-enous paramagnetic contrast agent, a decrease in itsconcentration is reflected as an increase in signal in-tensity on GRE images.

To perform functional MR imaging, a high-resolu-tion, three-dimensional, spoiled gradient-recalled ac-quisition in the steady state (steady-precession GRE)T1-weighted whole-brain study is initially performed.Then, for collection of functional MR imaging data,GRE echo-planar imaging is performed. The magneticevoked responses to the stimulated tasks (eg, workingmemory) are subsequently coregistered with the high-resolution MR images. Functional MR imaging researchin traumatic brain injury has centered primarily on theassessment of deficits after mild traumatic brain injury,but functional MR imaging also offers promise in theunderstanding of the ability of the brain to reorganizeafter injury (14). Brain SPECT, PET, and functional MRimaging show promise, but many of the currently avail-able data rely on small case series or case reports. Notethat disturbances in cerebral blood flow can also bemeasured with CT perfusion or MR perfusion imaging.

EXTRAAXIAL INJURIES

Fractures

This is CT territory. Although the primary role of CTin evaluating acute traumatic brain injury is to quicklyidentify a neurosurgical lesion, CT images optimally

demonstrate fractures of the calvaria, skull base, andfacial skeleton. Newer three-dimensional technology,which allows remarkable display of complex fractures,is now expeditious and user friendly and is becomingmore widely available.

In addition to routine scrutiny of the axial sections,vertex and horizontal calvarial fractures should besought on the sagittal scout image, where they mayoccasionally be better depicted. The presence of fluidlevels in the sphenoid sinuses or fluid within the mas-toid air cells and/or middle ear cavity should alert oneto the possibility of a skull base or temporal bonefracture. An air-fluid level in a maxillary sinus raisesconcern about an orbital floor fracture (ie, facialtrauma), although such an air-fluid level is also fre-quently noted in intubated patients as a result of re-tained sinus secretions. The 5-mm-thick sectionsthrough the brain for a routine head scan are insuffi-cient to evaluate these areas further, and additionalaxial 1-mm sections through the skull base or orbitsmay be helpful. Indeed, 60% of temporal bone frac-tures may be missed on routine CT images (31). Air inthe ipsilateral temporomandibular joint is a helpfulsecondary manifestation of temporal bone trauma(32). All fractures should be assessed for the presenceof a scalp laceration and radiopaque foreign bodies,both of which increase the likelihood of infection.

On the lowest axial section of a routine brain scan,the first cervical vertebra is often identified, and its re-lation to the dens should be noted. It is unusual, how-ever, to see any further portion of the cervical spine,although the lateral scout image may again provideclues to cervical injury (Fig 2). For complete rapidevaluation of the cervical spine when radiographs areabnormal or insufficient, 1-mm helical sections shouldbe obtained, ideally at the time of the brain study, andreformatted in the sagittal and coronal planes. In high-risk patients, radiographs are becoming obsolete.

Figure 2. Role of scrutinizing the scoutview. This 62-year-old woman with a priorhistory of traumatic brain injury came to theemergency department following a seizure.(a) Axial CT image at admission demon-strates left frontal posttraumatic encephalo-malacia subjacent to a craniotomy. No acuteintracranial abnormality is seen. (b) Lateralscout view, however, reveals a displacedfracture of cervical vertebra C2 (circle).

Gea

n e

t al

22

Epidural Hematoma

The epidural hematoma occurs in the potentialspace located between the dura and the inner table ofthe skull (Fig 3) (4). More than 75% occur in the re-gion of the temporal squamosa, and a vast majority ofepidural hematomas are associated with an underly-ing skull fracture (6). The characteristic hyperattenu-ating biconvex contour of an epidural hematoma re-sults from tense distention of the epidural space witharterial blood, typically from the middle meningealartery. The venous epidural hematoma, as an excep-tion to this, occurs at three classic sites: (a) in the pos-terior fossa from rupture of the torcular or the trans-verse sinus, (b) in the middle cranial fossa from dis-ruption of the sphenoparietal sinus, and (c) at thevertex, with hemorrhage from the superior sagittal si-nus (33). The latter lesion can be difficult to diagnosewith axial CT images, although it can be readily con-firmed with direct coronal images or coronal recon-structions of the axial data. Unlike the arterial epidu-ral hematoma, the venous epidural hematoma rarelyexpands beyond the initial injury because of the lowerpressure generated by venous extravasation.

An important CT finding suggestive of rapid expan-sion of an arterial epidural hematoma is the presenceof low-attenuation areas within an otherwise hyper-attenuating hematoma. This appearance has been re-ferred to as “the swirl sign” and is thought to representactive bleeding (Fig 4) (34,35). A heterogeneous epi-dural hematoma is an ominous finding that shouldbe followed closely.

Although differentiation from the subdural he-matoma usually is straightforward, several imagingfeatures are useful in making this distinction. First,because the dura is tightly adherent at sutures, it isuncommon for the epidural hematoma to cross a su-ture line (with the exception of the sagittal suture).Second, the epidural hematoma can cross the mid-line, in contrast to the subdural hematoma, which islimited by the falx. Third, the epidural hematoma isseen as a sharply localized lentiform mass, with thedisplaced dura seen on the MR image as an intactlow-signal-intensity line between the brain and theepidural mass (Fig 5). This hypointense line is avaluable finding when the shape and location of theextraaxial collection are inconclusive. Fourth, theepidural hematoma can extend from the supratento-rial to the infratentorial space, whereas the subduralhematoma is limited by the tentorium. Fifth, 99% ofepidural hematomas are located at the coup site,whereas the subdural hematoma is usually found atthe contrecoup site. Finally, because the venous si-nuses are composed of both dural layers, their dis-placement confirms an epidural mass.

The clinical manifestation of the epidural hema-toma is governed by the mass effect produced and by

the associated neuronal injuries. Brain injury, how-ever, is much less common with an epidural hema-toma than with a subdural hematoma. The classicmanifestation is that of a patient who experiences alucid conscious interval that is soon followed by neu-rologic deterioration. The lucid interval is attributedto the absence of underlying brain injury, with subse-quent enlargement of the epidural hematoma result-ing in progressive neurologic decline. However, thisclassic manifestation is seen in only 25% of the pa-tients (36,37).

Subdural Hematoma

The subdural hematoma is a serosanguineous fluidcollection located between the arachnoid mater and thedura. Subdural hematoma is seen in 10%–20% of pa-tients with closed head trauma and is the most com-mon operable intracranial hematoma (38). The subdu-ral hematoma is usually due to laceration of the bridg-ing cortical veins (Fig 6). An increased incidence ofsubdural hematoma is seen in elderly patients becausecerebral atrophy allows increased relative motion be-tween the brain parenchyma and calvaria and becausethe atrophic brain is less capable of tamponade of asmall subdural hematoma. The degree of brain distor-tion in these individuals, however, may be relatively mi-nor because of the abundant extracerebral space (39).

Rapid decompression of obstructive hydrocephaluscan also result in a subdural hematoma if the brainsurface recedes from the dura faster than the paren-chyma reexpands after having been compressed by thedistended ventricles. Other causes of a subdural he-matoma include injury to pial vessels, pacchioniangranulations (arachnoid granulations), or the greatveins. The acute subdural hematoma is associatedwith a skull fracture in less than 50% of the cases andis typically a contrecoup lesion. Most subdural hemato-

Figure 3. Subdural hematoma versus epidural hematoma.Coronal schematic through the level of superior sagittal sinus(S) illustrates the relationship of a subdural hematoma and anepidural hematoma to normal meningeal anatomic structures.Note that dura is a bilayered structure and that the layers splitto form dural venous sinuses and falx. The epidural hematoma(EDH) is located above the outer dural layer (ie, periosteum).The subdural hematoma (SDH) is located beneath the inner du-ral layer. (Reprinted, with permission, from reference 4.)

Traum

atic Brain In

jury

23

in attenuation, and is associated with greater mass ef-fect than expected for the size of the collection (Fig7). The heterogeneity represents recent unclottedhemorrhage, anemia, a paucity of hemoglobin withinthe hematoma, or admixing with cerebrospinal fluidfrom torn arachnoid mater (4,40).

In contrast to a heterogeneous acute subdural he-matoma, a heterogeneous chronic subdural hema-toma shows fluid-fluid levels, septa, and loculations,indicating its chronicity and its fragile vascular liningmembrane. The frequent absence of associated intra-axial injury is an additional imaging finding thathelps differentiate the heterogeneous acute subduralhematoma from repeat bleeding into a preexistingchronic subdural hematoma.

As mentioned previously, the subdural hematoma(unlike the epidural hematoma) crosses calvarial su-tures and can extend along the tentorium and the falxcerebri. The subdural hematoma does not, however,typically travel from the supratentorial space into theposterior fossa. A vast majority of subdural hemato-mas are supratentorial in location. The posterior fossasubdural hematoma is uncommon in adults, prob-ably because of the greater restriction of rotationalmovement of the cerebellum and diminished shearingstress on superficial veins.

Although the shape of the typical acute subdural he-matoma tends to be crescentic, the formation of duraladhesions (eg, following trauma or infection) can re-sult in a more convex collection, thus mimicking theepidural hematoma. This is particularly challenging inthe elderly patient.

There are two relatively common pitfalls in the CTevaluation of the subdural hematoma (Fig 8). The firstpitfall occurs with a thin-convexity hematoma, whichcan be difficult to appreciate adjacent to the hyper-attenuating skull unless the CT image is viewed withwide windows (so-called blood windows). The intro-duction of a picture archiving and communications

Figure 4. Active bleeding within an epidu-ral hematoma. (a) Admission axial CT im-age and (b) 3-hour follow-up axial CT imagedemonstrate marked interval enlargement ofa right temporal epidural hematoma. Notethe intrinsic hypoattenuation that representsactive bleeding (arrow); the high-attenuationarea represents clotted blood. Also note themarked increase in scalp soft-tissue swell-ing (double-headed arrow) at the coup site.

Figure 5.Venous epiduralhematoma (dis-placed dura). Inter-mediate-weightedaxial MR imagedemonstrates athin black line (ar-row) at the medialmargin of thehyperintense col-lection. This linerepresents the dis-placed dura, andits presence con-firms the epidurallocation of lesion.Current epiduralhematoma wassecondary to inter-ruption of the righttransverse sinus. Whether venous or arterial, all epidural he-matomas show displaced dura at MR imaging. A contrecoup leftorbitofrontal contusion (circle) is also noted.

mas are semisolid within 1–10 days and thus require acraniotomy, rather than burr hole evacuation. Withtime, the subdural hematoma organizes via activated fi-broblasts and blood vessels that invade the hematomafrom the dura. These vessels are fragile and are prone toepisodes of repeat bleeding (hence, the problem of the“chronic recurrent subdural hematoma”).

The CT appearance of the typical acute subduralhematoma consists of a hyperattenuating, homoge-neous, crescentic contrecoup lesion that frequentlytracks along the entire hemisphere. The degree ofmass effect seen in association with the subdural he-matoma often appears excessive for the size of thehematoma because of the presence of underlying pa-renchymal injury. If appropriate midline shift is notseen, then a contralateral mass lesion should be sus-pected. An “atypical” acute subdural hematomatends to be less crescentic in shape, is heterogeneous

Gea

n e

t al

24

system allows the film reviewer to adjust the windowsettings readily. The second pitfall in the CT evaluationof the subdural hematoma is subacute bleeding thathas become isoattenuating with adjacent brain. Recog-nition of secondary signs, such as displacement of graymatter, white matter buckling, mass effect, and midlineshift, is helpful, although it can be difficult when thesubdural hematoma is bilateral. Contrast opacificationof the displaced cortical veins (or MR imaging) canprovide confirmatory evidence of the isoattenuatingsubdural hematoma (41–43).

The chronic subdural hematoma is frequently bilat-eral and, in the absence of recurrent hemorrhage, mayappear uniformly low in attenuation, thus mimickinga subdural hygroma (Fig 9) or even cerebral atrophy.The presence of mass effect, however, should excludethe latter.

Subdural Hygroma

The traumatic subdural hygroma is thought to arisefrom cerebrospinal fluid leakage and a “one-way” rentin the arachnoid mater. The traumatic subdural hy-groma generally develops more than 3 days after trau-matic brain injury. These fluid collections have cere-brospinal fluid attenuation on CT images and may beindistinguishable from a chronic subdural hematoma(Fig 9). A comparison study can be invaluable in thesecases. The distinction can sometimes be made by evalu-ating the attenuation of the collection (Hounsfieldunits), which is equivalent to that of cerebrospinalfluid for the subdural hygroma. MR imaging is capableof clarifying any ongoing confusion because it is moresensitive than CT to the presence of blood products in asubdural hematoma. On MR images, the subdural hy-groma follows cerebrospinal fluid in signal intensity,whereas a chronic subdural hematoma is of higher sig-nal intensity with all pulse sequences.

The signs and symptoms of a subdural hygromaare sometimes indistinguishable from those of a sub-acute subdural hematoma. Most patients, however,are asymptomatic, and their condition is managedconservatively (44). In 85% of the patients withclinical symptoms referable to a subdural hygroma,the condition responds to burr hole evacuation.

Subarachnoid Hemorrhage

The most common cause of subarachnoid hemor-rhage is trauma. The blood can arise from direct pialinjury, extension from an underlying parenchymalcontusion, or contiguous extension of intraventricu-lar hemorrhage. The interpeduncular cistern andsylvian fissures are two common sites for the accu-mulation of traumatic subarachnoid hemorrhage(Fig 10). Therefore, establishing a clinical history canbe crucial to avoid mistaking traumatic subarach-noid hemorrhage for a ruptured cerebral aneurysm(45,46). Although uncommon, aneurysmal rupture

Figure 6. Bridg-ing cortical veins.(a) Coronal and(b) axial contrast-enhanced T1-weighted MR im-ages demonstrateenhancement ofseveral veins (ar-rows) traversingthe subdural andsubarachnoidspace. These ves-sels are the onesthat are torn duringsudden accelera-tion and decelera-tion, thus leadingto subdural he-matoma.

Figure 7. Acute"atypical" subduralhematoma. Thisleft holohemi-spheric subduralhematoma istermed atypicalbecause of its het-erogeneity, its dis-proportionatemass effect for itssize, and itsslightly convexshape. Atypicalsubdural hema-toma has a par-ticularly ominousprognosis. An-other highly un-usual finding inthis case is braininjury at the coupsite (note left-sided scalp soft-tissue swelling). Extensive sub-arachnoid blood overlying left hemisphere is an additional poorprognostic sign.

Traum

atic Brain In

jury

25

may cause the trauma, resulting in imaging featuresof both conditions.

On CT images, acute subarachnoid hemorrhage isseen as abnormal linear areas of high attenuation inter-digitating between sulci and fissures. As is the case with

the majority of traumatic lesions, the largest amount ofsubarachnoid hemorrhage is usually located at the con-trecoup site. With time, the blood becomes isoatten-uating with brain and ultimately resolves. In the sub-acute and chronic phase, MR imaging is still able to

Figure 8. Two potential pitfalls in CT evalu-ation of subdural hematoma. (a) Coronal T1-weighted MR image demonstrates thin linearhyperintense lesions (arrows) beneath leftoccipital lobe, as well as lateral to right cer-ebellar hemisphere, consistent with smallsubacute subdural hemorrhages. Because ofthe small size of the collection and obscura-tion by beam-hardening artifact of the calva-ria, the corresponding CT study was inter-preted as normal. (b) Nonenhanced axial CTimage of a different patient shows bilateralmedial displacement of cortical surface(double-headed arrows) and "white matterbuckling," consistent with bilateral isodensesubdural hematoma.

Figure 9. Acute subdural hygroma.(a) Axial CT image on admission shows leftparieto-occipital scalp soft-tissue swelling(arrow) and no intracranial abnormality.(b) Three-day follow-up axial CT imagedemonstrates interval development ofsmall bifrontal low-attenuation subduralcollections. If prior CT images were notavailable, the appearance could be con-fused with chronic subdural hematomas.

Figure 10. Traumatic subarachnoid hem-orrhage (two common locations). (a) AxialCT image shows small amount of hyper-attenuating blood (arrow) within the interpe-duncular fossa. Brain is otherwise normal.(b) Axial CT image from another patientshows subarachnoid hemorrhage within theright sylvian fissure. Note characteristiccontrecoup location of subarachnoid hemor-rhage, identified opposite the scalp injury(arrow). In these two cases, the cause ofthe hemorrhage is likely secondary toshearing of perimesencephalic and insularveins, respectively.

Gea

n e

t al

26

depict subtle residual blood, particularly with FLAIRand GRE sequences (Fig 11). Subarachnoid hemor-rhage is toxic to the underlying cortex and can alsointerfere with normal cerebrospinal fluid resorptionat the level of the pacchionian granulations, thusleading to communicating hydrocephalus.

Intraventricular Hemorrhage

Traumatic intraventricular hemorrhage can occur byone of three methods: (a) contiguous extension froma parenchymal hematoma, (b) shearing of subependy-mal veins that line the ventricular cavities, or (c) retro-grade reflux of subarachnoid hemorrhage through theforamina of the fourth ventricle. Traumatic intraven-tricular hemorrhage may be isolated, but it is usuallyassociated with superficial contusions and subarach-noid hemorrhage. Subtle intraventricular hemorrhagecan be detected by the appearance of a fluid-fluid levellayering dependently within the occipital horns of thelateral ventricles (so-called hematocrit effect), becausefibrinolytic activators within cerebrospinal fluid inhibitclotting (38). In some cases, the choroid plexus may actas a nidus for the blood to clot and form a ventricularcast or tumefactive blood clot (Fig 12). In the absenceof recurrent hemorrhage, the blood rarely persists formore than 1–2 weeks. Large amounts of intraventricu-lar blood may impede cerebrospinal fluid flow and re-sult in noncommunicating hydrocephalus.

INTRAAXIAL INJURIES

Traumatic Dysautoregulation

With normal circumstances, the amount of cerebralblood flow to the brain remains constant despite vac-illations in blood pressure. This ability to regulateblood flow, which is termed autoregulation, stemsfrom the unique capacity of the cerebral vasculature todilate in response to a decrease in blood pressure andconstrict in response to an increase in blood pressure.The loss of this ability, or dysautoregulation, is a well-recognized sequela of traumatic brain injury that isparticularly common in younger individuals. In theearly stages of dysautoregulation, the hyperemic swell-ing appears on CT and MR images as ill-defined focal

or diffuse sulcal effacement with preservation of thegray-white differentiation (Fig 13). With increasing se-verity, the hyperemia may progress to cerebral edema.

Cortical Contusion

Cortical contusions are hemorrhagic parenchymal“bruises.” They are peripheral lesions, involving the

Figure 11. Sub-acute traumaticsubarachnoidhemorrhage. Coro-nal FLAIR MR im-age demonstrateshorizontal linearhyperintensity (ar-row) within severalleft temporal-oc-cipital sulci. Corre-sponding CT studywas normal.

Figure 12. Iso-lated intraventricu-lar hemorrhage.(a) Axial CT imagedemonstrateshyperattenuatingblood layeringwithin the occipitalhorn of the left lat-eral ventricle, pro-ducing the so-called hematocriteffect (arrow). Im-age is otherwisenormal. (b) T1-weighted axial MRimage of a differ-ent patient is re-markable for smallround hyperintensenodules (arrows)located within an-terior right tempo-ral horn and de-pendently withinleft occipital horn,consistent withtumefactive clots.In both of thesecases, intraven-tricular hemor-rhage is likely sec-ondary to shearingof subependymalveins.

Figure 13. Dys-autoregulation.Axial CT imageshows asymmetryof the sylvian fis-sure and temporalsulci, with preser-vation of gray-white differentia-tion. In addition tohyperemic swellingof right temporallobe, the differen-tial diagnosiswould includeisodense sub-arachnoid hemor-rhage overlyingright hemisphere, meningitis, a small right subdural hematoma,and asymmetric atrophy involving left hemisphere.

Traum

atic Brain In

jury

27

gyral crests, particularly those in contact with irregularcontours of the skull (eg, the orbital roof, petrousridge, and sphenoid ridge) (38,40). This superficial lo-

cation makes detection with CT difficult, particularlywhen patient motion occurs and there is streak arti-fact. MR imaging offers an important advantage in thedetection of these often subtle injuries because the cal-varia does not distort the signal. MR imaging is alsocapable of providing images in multiple planes, obvi-ating partial volume artifacts from one plane alone.

On T2-weighted MR images, contusions are seen asareas of increased signal intensity (depending on theextent of hemorrhage) and are particularly conspicuouswith the FLAIR sequence. Although currently the mostsensitive MR techniques for blood products are multi-planar gradient-recalled imaging and susceptibility-weighted MR imaging, susceptibility artifact from theskull limits evaluation of surface contusions. Withtime, the lesion shrinks into a gliotic scar. An old con-tusion is seen as a wedge-shaped area of peripheral en-cephalomalacia, with the apex of the wedge pointingcentrally and the broad base facing the irregular surfaceof the skull. In the chronic stage, this triangular shapecan resemble a remote ischemic infarct (Fig 14).

Cerebral Hematoma

Unlike the cerebral contusion, the cerebral he-matoma is located deeper in the brain. In the acutesetting, the hematoma is hyperattenuating on CT im-ages. Within days, a peripheral rim of low attenuationis seen, consistent with edema and pressure necrosis.In the subacute phase, ring enhancement can be notedwith either CT or MR imaging because of the prolif-eration of new capillaries lacking a complete blood-brain barrier. Indeed, in the absence of prior studiesor an accurate clinical history, it may not be possibleto differentiate a hematoma with ring enhancementfrom an abscess, neoplasm, or infarct. In the chronicphase, a smaller area of nonenhancing encephaloma-lacia results, with compensatory dilatation of the ipsi-lateral ventricle and sulci. On CT images, this appear-ance is nonspecific, but the presence of blood prod-ucts can be detectable with MR imaging for years.

The “coup-contrecoup” mechanism refers to thefact that the moving skull comes to an abrupt stopwhile the brain continues to move for another briefmoment (Figs 15, 16) (4,47–50). The portion of the

Figure 14. Cortical contusion. (a) Para-sagittal T1-weighted MR image demon-strates multiple wedge-shaped areas ofhyperintensity (arrows), consistent with sub-acute cortical contusions. Lesions involve thesurface of brain. They are wedge-shaped,with their base abutting calvarial surface andtheir apex pointing centrally. (b) T2-weightedaxial MR image shows well-defined wedge-shaped areas of hyperintensity (arrow) withinleft temporal cortex, consistent with remotetrauma (posttraumatic encephalomalacia).Without a proper clinical history, lesion couldbe mistaken for remote ischemic infarction.

Figure 15. Coup-contrecoup mechanism. On impact, a de-crease in parenchymal pressure occurs within the frontal lobesas they are transiently displaced away from skull. Frontal bridg-ing cortical veins are also lacerated at this time. In contrast, theoccipital lobes experience momentary increase in pressure asthey are thrust against the coup site. Negative pressure gradi-ents are toxic to the brain. Resultant injury classically consistsof scalp swelling, skull fracture, and epidural hematoma at thecoup site, with a contrecoup subdural hematoma and intraaxialhemorrhage. (Reprinted, with permission, from reference 4.)

Figure 16. Coup-contrecoup injury.Axial CT imageshows epidural he-matoma and scalpinjury at the coupsite and subduralhematoma andintraaxial injury atthe contrecoupsite. Extent of mid-line shift is lessbecause of thebalanced mass ef-fect from the twolesions.

Gea

n e

t al

28

brain opposite the impact site initially pulls away fromthe dura but on recoil strikes the dura with force. Injuryat the impact site is termed the coup injury, whereasthat on the opposite side is termed the contrecoup in-jury. The coup site is identified by the presence of askull fracture, epidural hematoma, or scalp injury. Al-though both coup and contrecoup lesions can result inhemorrhage, it is far more common at the contrecoupsite. Interestingly, contrecoup lesions are rarely seen inthe cerebellum or occipital lobes after a frontal impactbecause of the thick smooth inner surface of the occipi-tal bone and because the falx and tentorium act to sta-bilize the adjacent brain parenchyma.

Occasionally, patients with head injuries develop adelayed hematoma in an area previously shown to benonhemorrhagic on CT or MR images (Fig 17) (51–53). Such hematomas tend to be lobar, are frequentlymultiple, and often occur in areas that demonstratedcontusion on initial images. Most delayed hematomasoccur within 2–4 days of injury and are associatedwith a poor prognosis. The pathogenesis of these le-sions is controversial, but it probably is related to thefact that ischemic tissue is extremely vulnerable toreperfusion hemorrhage. Possible causes include va-sospasm with subsequent vasodilation, hypotensionwith subsequent hypertension, and a preexisting oracquired coagulopathy.

The attenuation of acute hemorrhage on CT imagesis related to the globin moiety of hemoglobin (notto the iron component). Acute blood normally hasan attenuation value of 50–70 HU, whereas brain pa-renchyma measures 20–30 HU. Acute hematomas inanemic patients (hemoglobin level, <11 mg/dL) maythus appear isoattenuating with brain parenchyma.The attenuation of an acute clot may increase in thefirst few days as the clot retracts. Subsequent proteoly-sis results in a decrease in attenuation. The MR imag-ing features of hemorrhage are influenced by the fol-

lowing: (a) location (subarachnoid or subdural orintraaxial), (b) field strength (magnetic susceptibilityis proportional to field strength [2]), (c) pulse se-quence, (d) lesion size, (e) clot retraction, (f) redblood cell integrity, (g) the presence or absence ofcontinued bleeding, (h) hemoglobin oxygenationstate, (i) local tissue pH, and (j) oxygen tension. Thedetails of these factors are complex and beyond thescope of this chapter. Cerebral edema, seen as a pe-

Figure 17. Delayed traumatic hematoma.(a) Admission axial CT image reveals ef-facement of the right occipital horn but nodefinite intraaxial hemorrhage. (b) Six-hourfollow-up CT image shows interval develop-ment of a large right temporal hematomawith a fluid-fluid level (arrow).

Figure 18. Dif-fuse axonal injury.(a) ConventionalT2-weighted axialspin-echo MR im-age demonstratesseveral subtlepunctate foci ofhypointensity (ar-rows) within thefrontal subcorticalwhite matter. (b) Coronal GREMR image of samepatient betterdemonstrates theextent of signalabnormality. Thisloss of signal isdue to the ferro-magnetic effect ofblood productssuch as hemosid-erin and ferritin.

Traum

atic Brain In

jury

29

ripheral zone of low attenuation surrounding thehemorrhage, appears at about 8 hours and is maximal3–5 days after the initial injury. The degree of edemais a function of both the extent of the initial injuryand the state of hydration of the patient.

Diffuse Axonal Injury

Although diffuse axonal injury is a nonsurgical in-jury, its presence has long-term implications with re-spect to patient prognosis. It is well accepted that CTremains limited in the detection of diffuse axonal in-jury, with images frequently being discordant with theclinical status of the patient (14,54–58). In the acutephase, discrete small (usually <1-cm) foci of hypo-attenuation may be depicted in the white matter at thegray-white junction. Acute hemorrhagic shear injuriesare more easily identified, but brainstem and poste-rior fossa diffuse axonal injury often remains elusive

because of beam-hardening artifact. In addition, thehemorrhage may resolve quickly and not be visible onthe follow-up images.

Because of the relatively poor sensitivity of CT in thedetection of diffuse axonal injury, MR imaging may behelpful in the first 2 weeks to better evaluate the ex-tent of injury. GRE and susceptibility-weighted MRimages are particularly helpful because of their en-hanced sensitivity to the presence of blood products(Fig 18) (16,59,60). It is important to remember thatfast spin-echo imaging is less sensitive to diffuse ax-onal injury (and to hemorrhage, in general) than isconventional spin-echo MR imaging.

The most common shearing lesions are seen in thefrontal parasagittal white matter. With increasingshearing force, the corpus callosum, typically the sple-nium, is injured (Figs 19, 20) (61). The most severeshearing forces result in dorsolateral brainstem injury

Figure 19. Diffuse axonal injury of the cor-pus callosum. (a) T1-weighted midsagittalMR image demonstrates a slightly prominentsplenium of the corpus callosum with mini-mal hypointensity (arrow). (b) Correspond-ing T2-weighted midsagittal MR imageclearly shows the central splenial signal ab-normality, with subtle sparing of the callosalsurface (arrow).

Figure 20. There are currently a number ofMR pulse sequences that are used to iden-tify diffuse axonal injury (DAI). (See text foradvantages and disadvantages.) ADC = ap-parent diffusion coefficient map, DTI = diffu-sion-tensor imaging, DWI = diffusion-weighted imaging, MPGR = multiplanar gra-dient-recalled imaging, SWI = susceptibility-weighted imaging.

Gea

n e

t al

30

(ie, “the deeper the lesion location, the more severethe injury”). These three locations (parasagittal whitematter, corpus callosum, and dorsolateral brainstem)are called the shear injury triad. Acute nonhemorrhagiclesions of diffuse axonal injury are bright on diffu-sion-weighted images, with a corresponding decreasein apparent diffusion coefficient (19). As mentionedpreviously, diffusion-tensor imaging is being investi-gated as a means to detect more subtle axonal injurythat can reflect dysfunction of white matter tracts. In-jury disrupts the preferential mobility (anisotropy) ofwater along the fiber tracts, and the findings can be dis-played quantitatively in the form of fractional anisot-ropy or with visual depiction on color tensor maps.Findings from diffusion-tensor imaging may be abnor-mal with normal MR and CT images. After about 3weeks, diffuse axonal injury is often associated withatrophic enlargement of the ventricles and sulci.

Penetrating Injuries

More than 80% of the gunshot wounds to the headpenetrate the skull, and most of these patients die.Brainstem involvement is uniformly fatal, as is bihemi-spheric injury unless the injury is limited to the cerebralsurface. Forty percent of the surviving patients sufferposttraumatic epilepsy. The bullet trajectory can be de-termined by examining the calvaria and adjacent brainparenchyma (62–64). At the entry site, the inner tableof the skull is beveled, whereas at the exit site (usuallylarger), the outer table of the skull is beveled. The exitsite also tends to show greater bone destruction. High-velocity bullets cause more extensive direct tissue lac-eration and can result in severe contusion injuries at re-mote locations because of transmission of radial shockwaves. Penetrating and perforating injuries may becomplicated by fragments of bone or scalp within thebrain. It is the organic material, rather than the inor-ganic metal fragments, that accounts for the majority ofinfections after penetrating injuries.

Traumatic Brainstem Injury

Evaluating the brainstem and posterior fossa withCT is particularly difficult because of beam-hardeningartifact from bone. Indeed, CT depicts only 20% ofthe acute brainstem injuries that are identified subse-quently with MR imaging. Nevertheless, in the acutesetting, CT remains the imaging modality of choicebecause the brainstem contusion and shearing inju-ries are managed conservatively. MR imaging may bewarranted in the subacute setting to investigate unex-plained neurologic deficits and to determine prognosis.In addition, the ability to better image these areas withMR has furthered our understanding of traumatic braininjury. Primary injuries to the brainstem have a predi-lection for the dorsolateral mesencephalon and includedirect contusions and shear injuries (Fig 21) (65–70).Contusions may be isolated and typically result from

the impact of the brainstem against the tentorium.Contusions always extend to the brainstem surface.

Shearing injuries may occasionally be difficult todistinguish from contusions, but shearing injuries of-ten do not extend to the brainstem surface and are al-most always associated with supratentorial diffuse ax-onal injury. The secondary types of brainstem injuryinclude those resulting from hypoxic-ischemic dam-age or from abrupt downward herniation of thebrainstem (Duret hemorrhage) (Fig 22). In the latterinjury, brainstem descent causes stretching and shear-ing of perforating basilar arterial branches, which pro-duces hemorrhage within the brainstem. The Durethemorrhage also can arise either from vessel wall rup-ture caused by hypoxic damage or from venous infarc-tion. In contrast to the brainstem shearing injury andthe direct brainstem contusion, both of which have apredilection for the dorsolateral brainstem, the Durethemorrhage typically is seen in the central pons ormidbrain. Duret hemorrhage is associated with an ex-tremely poor prognosis. Uncommon causes of trau-matic brainstem injury include the pontomedullaryrent and hypoxic-ischemic injury.

In summary, although CT remains the mainstay forassessment of patients with acute traumatic brain in-jury, MR imaging has a greater sensitivity for the detec-tion of pathologic findings. MR imaging is currentlyused primarily for patients with unexplained brain dys-function after traumatic brain injury. The use of special-ized MR imaging techniques, such as diffusion-tensorimaging, diffusion-weighted imaging, susceptibility-weighted imaging, and MR spectroscopy, and func-tional studies such as functional MR imaging, PET, andSPECT may not yet alter immediate surgical care, butthese techniques do increase our understanding of thepathophysiology of trauma and of long-term neuro-logic disability. This, in turn, may allow more accurateprediction of patient outcome, may play a role in clini-cal drug trials, and may alter the direction of treatmentand rehabilitation of the patient with head injury.

Figure 21. Brain-stem contusion.Axial CT imagethrough the mid-brain shows dor-solateral hyper-attenuation withinthe left tectalplate, consistentwith a focal contu-sion secondary todirect impact withthe rigid duralmargin of the ten-torium. Brain isotherwise unre-markable.

Traum

atic Brain In

jury

31

References1. MacDorman MF, Minimo AM, Strobino DM, Guyer B. An-

nual summary of vital statistics: 2001. Pediatrics 2002; 110:1037–1052.

2. Lewin ICF. The cost of disorders of the brain. Washington,DC: National Foundation of the Brain, 1992.

3. Sato Y, Yuh WT, Smith WL, Alexander RC, Kao SC,Ellerbroek CJ. Head injury in child abuse: evaluation withMR imaging. Radiology 1989; 173:653–657.

4. Gean AD. Imaging of head trauma. New York, NY: Lippin-cott, Williams & Wilkins, 1994.

5. Sosin DM, Sniezek JE, Thurman DJ. Incidence of mild andmoderate brain injury in the United States. Brain Inj 1996;10:47–54.

6. Zee CS, Go JL. CT of head trauma. Neuroimaging Clin NAm 1998; 8:525–539.

7. Jeret JS, Mandell M, Anziska B, et al. Clinical predictors ofabnormality disclosed by computed tomography after mildhead trauma. Neurosurgery 1993; 32:9–15.

8. Miller EC, Holmes JF, Derlet RW. Utilizing clinical factors toreduce head CT scan ordering for minor head trauma pa-tients. J Emerg Med 1997; 15:453–457.

9. Haydel MJ, Preston CA, Mills TJ, Luber S, Blaudeau E,DeBlieux PM. Indications for computed tomography in pa-tients with minor head injury. N Engl J Med 2000; 343:100–105.

10. Ashikaga R, Araki Y, Ishida O. MRI of head injury usingFLAIR. Neuroradiology 1997; 39:239–242.

11. Singer MB, Atlas SW, Drayer BP. Subarachnoid space dis-ease: diagnosis with fluid-attenuated inversion-recovery MRimaging and comparison with gadolinium-enhanced spin-echo MR imaging—blinded reader study. Radiology 1998;208:417–422.

12. Frigon C, Jardine DS, Weinberger E, Heckbert SR, ShawDW. Fraction of inspired oxygen in relation to cerebrospinalfluid hyperintensity on FLAIR MR imaging of the brain inchildren and young adults undergoing anesthesia. AJR AmJ Roentgenol 2002; 179:791–796.

13. Kuzma BB, Goodman JM. Improved identification of axonalshear injuries with gradient echo MR technique. SurgNeurol 2000; 53:400–402.

14. Hammoud DA, Wasserman BA. Diffuse axonal injuries:pathophysiology and imaging. Neuroimaging Clin N Am2002; 12:205–216.

15. Reichenbach JR, Venkatesan R, Schillinger DJ, Kido DK,Haacke EM. Small vessels in the human brain: MR venog-raphy with deoxyhemoglobin as an intrinsic contrast agent.Radiology 1997; 204:272–277.

16. Tong KA, Ashwal S, Holshouser BA, et al. Hemorrhagicshearing lesions in children and adolescents with posttrau-matic diffuse axonal injury: improved detection and initialresults. Radiology 2003; 227:332–339.

17. Melhem ER, ed. Diffusion imaging. Neuroimaging Clin NAm 2002; 12(1) (special issue).

18. Huisman TA, Sorensen AG, Hergan K, Gonzalez RG,Schafer PW. Diffusion-weighted imaging for the evaluationof diffuse axonal injury in closed head injury. J Comput As-sist Tomogr 2003; 27:5–11.

19. Rugg-Gunn FJ, Symms MR, Barker GJ, Greenwood R,Duncan JS. Diffusion imaging shows abnormalities afterblunt head trauma when conventional magnetic resonanceimaging is normal. J Neurol Neurosurg Psychiatry 2001; 70:530–533.

20. Liu AY, Maldjian JA, Bagley LJ, Sinson GP, Grossman RI.Traumatic brain injury: diffusion-weighted MR imaging find-ings. AJNR Am J Neuroradiol 1999; 20:1636–1641.

21. Arfanakis K, Haughton VM, Carew JD, Rogers BP,Dempsey RJ, Meyerand ME. Diffusion tensor MR imagingin diffuse axonal injury. AJNR Am J Neuroradiol 2002; 23:794–802.

22. Castillo M, ed. Proton MR spectroscopy of the brain.Neuroimaging Clin N Am 1998; 8(4) (special issue).

23. Condon B, Oluoch-Olunya D, Hadley D, Teasdale G,Wagstaff A. Early magnetic resonance spectroscopy ofacute head injury: four cases. J Neurotrauma 1998; 15:563–571.

24. Garnett MR, Blamire AM, Rajagopalan B, Styles P, Cadoux-Hudson TA. Evidence for cellular damage in normal-appear-ing white matter correlates with injury severity in patientsfollowing traumatic brain injury: a magnetic resonancespectroscopy study. Brain 2000; 123:1403–1409.

25. Garnett MR, Blamire AM, Corkill RG, Cadoux-Hudson TA,Rajagopalan B, Styles P. Early proton magnetic resonancespectroscopy in normal-appearing brain correlates with out-come in patients following traumatic brain injury. Brain2000; 123:2046–2054.

26. Garnett MR, Corkill RG, Blamire AM, et al. Altered cellularmetabolism following traumatic brain injury: a magneticresonance spectroscopy study. J Neurotrauma 2001; 18:231–240.

27. Camargo EE. Brain SPECT in neurology and psychiatry. JNucl Med 2001; 42:611–623.

28. Hofman PA, Stapert SZ, van Kroonenburgh MJ, Jolles J, deKruijk J, Wilmink JT. MR imaging, single-photon emissionCT, and neurocognitive performance after mild traumaticbrain injury. AJNR Am J Neuroradiol 2001; 22:441–449.

29. McAllister TW, Sparling MB, Flashman LA, Saykin AJ.Neuroimaging findings in mild traumatic brain injury. J ClinExp Neuropsychol 2001; 23:775–791.

30. Ogawa S, Lee TM, Nayak AS, Glynn P. Oxygenation-sensi-tive contrast in magnetic resonance imaging. Magn ResonMed 1990; 14:68–78.

31. Holland BA, Brant-Zawadzki M. High-resolution CT of tempo-ral bone trauma. AJR Am J Roentgenol 1984; 143:391–395.

Figure 22. Duret hemorrhage. (a) Preop-erative axial CT image demonstrates a righttemporal extraaxial collection (arrow) and anormal brainstem. Imaging manifestationsof downward herniation were noted on morecranially located CT images. (b) Postopera-tive axial CT image shows hemorrhagewithin the central pons.

Gea

n e

t al

32

32. Betz BW, Wiener MD. Air in the temporomandibular jointfossa: CT sign of temporal bone fracture. Radiology 1991;180:463–466.

33. Gean AD, Evans SSJ, McCormick V. The sphenoparietalvenous epidural hematoma: an indolent lesion with a char-acteristic imaging appearance (abstr). Presented at the 6thAnnual Meeting of the American Society of Emergency Ra-diology, Scottsdale, Ariz, March 1995.

34. Al-Nakshabandi NA. The swirl sign. Radiology 2001; 218:433.

35. Greenberg J, Cohen WA, Cooper PR. The "hyper-acute"extraaxial intracranial hematoma: computed tomographicfindings and clinical significance. Neurosurgery 1985; 17:48–56.

36. Cordobes F, Lobato RD, Rivas JJ, et al. Observation on 82patients with extradural hematomas. J Neurosurg 1981; 54:179–186.

37. Rivas JJ, Lobato RD, Sarabia R, et al. Extradural hema-toma: analysis of factors influencing the courses of 161 pa-tients. Neurosurgery 1988; 23:44–51.

38. Young RJ, Destian S. Imaging of traumatic intracranialhemorrhage. Neuroimaging Clin N Am 2002; 12:189–204.

39. Gennarelli TA, Thibault LE. Biomechanics of acute subduralhematoma. J Trauma 1982; 22:680–686.

40. Zee CS, Hovanessian A, Go JL, Kim PE. Imaging of se-quelae of head trauma. Neuroimaging Clin N Am 2002; 12:325–338.

41. Kaufman HH, Singer JM, Sadhu VK, et al. Isodense acutesubdural hematoma. J Comput Assist Tomogr 1980; 4:557–559.

42. Moller A, Ericson K. Computed tomography in isoatten-uating subdural hematomas. Radiology 1979; 130:149–152.

43. Hayman LA, Evans RA, Hinck VC. Rapid-high-dose con-trast computed tomography of isodense subdural hema-toma and cerebral swelling. Radiology 1979; 131:381–383.

44. Hoff J, Barnes E, Barnes B, et al. Traumatic subdural hy-groma. J Trauma 1973; 13:870–876.

45. Yeakley JW, Patchall LL, Lee KF. Interpeduncular fossasign: CT criterion of subarachnoid hemorrhage. Radiology1986; 158:699–700.

46. Gean AD, Sandu FS, McCormick V, Evans SJJ, Gruden JF.Congenital aneurysmal vs traumatic subarachnoid hemor-rhage: new insights. Presented at the 94th Annual Meetingof the American Roentgen Ray Society, New Orleans, La,May 1994.

47. Lindenberg R, Freytag E. The mechanism of cerebral con-tusions: a pathologic-anatomic study. Arch Pathol 1960; 69:440–469.

48. Holbourn AHS. Mechanics of head injuries. Lancet 1943; 2:438–441.

49. Dawson SL, Hirsch CS, Luas FV, et al. The contrecoupphenomenon: reappraisal of a classic problem. Hum Pathol1980; 11:155–166.

50. Gurdjian ES, Webster JE, Lissner HR. Observations on themechanism of brain concussion, contusion, and laceration.Surg Gynecol Obstet 1955; 101:680–690.

51. Zulch KJ. Delayed post-traumatic apoplexy. Neurosurg Rev1989; 12:252–253.

52. Nanassis K, Forwein RA, Karimi A, et al. Delayed post-trau-matic intracerebral bleeding. Neurosurg Rev 1989; 12:243–251.

53. Lipper MH, Kishore PR, Girevendulis AK, Miller JD, BeckerDP. Delayed intracranial hematoma in patients with severehead injury. Radiology 1979; 133:645–649.

54. Zimmerman RA, Bilaniuk LT, Genneralli T. Computed to-mography of shearing injuries of the cerebral white matter.Radiology 1978; 127:393–396.

55. Mittl RL, Grossman RI, Hiehle JF, et al. Prevalence of MRevidence of diffuse axonal injury in patients with mild headinjury and normal CT findings. AJNR Am J Neuroradiol1994; 15:1583–1589.

56. Gentry LR, Godersky JC, Thompson B. MR imaging ofhead trauma: review of the distribution and radiographicfeatures of traumatic lesions. AJR Am J Roentgenol 1988;150:663–672.

57. Gennarelli TA, Thibault LE, Adams JH, Graham DI, Thomp-son CJ, Marcincin RP. Diffuse axonal injury in traumaticcoma in the primate. Ann Neurol 1982; 12:564–574.

58. Adams JH, Graham DI, Murray LS, Scott G. Diffuse axonalinjury due to nonmissile head injury in humans: an analysisof 45 cases. Ann Neurol 1982; 12:557–563.

59. Tong KA. New MRI techniques for imaging of head trauma:DWI, MRS, SWI. Appl Radiol, July 2003; 29–35.

60. Choi RE, Smith RR, Edwards MK, et al. A comparison ofgradient-echo and spin-echo magnetic resonance in theevaluation of diffuse axonal injury. In: Proceedings of theAnnual Meeting of the American Society of Neuroradiology.Vol 56. Oak Brook, Ill: American Society of Neuroradiology,1991.

61. Gentry LR, Thompson B, Godersky JC. Trauma to the cor-pus callosum: MR features. AJNR Am J Neuroradiol 1988;9:1129–1138.

62. Wilson AJ. Gunshot injuries: what does a radiologist needto know? RadioGraphics 1999; 19:1358–1368.

63. Hollerman JJ, Fackler ML, Coldwell DM, Ben-Menachem Y.Gunshot wounds. I. Bullets, ballistics, and mechanisms ofinjury. AJR Am J Roentgenol 1990; 155:685–690.

64. Hollerman JJ, Fackler ML, Coldwell DM, Ben-Menachem Y.Gunshot wounds. II. Radiology. AJR Am J Roentgenol1990; 155:691–702.

65. Friede RL, Roessman U. The pathogenesis of secondarymidbrain hemorrhages. Neurology 1966; 16:1210–1216.

66. Gean AD, Evans SJJ, Patel M. Traumatic brainstem injury:incidence, classification, imaging manifestations and patho-genesis. Presented at the 7th annual meeting of the Ameri-can Society of Emergency Radiology, Orlando, Fla, April1996.

67. Gentry LR, Godersky JC, Thompson BH. Traumatic brainstem injury: MR imaging. Radiology 1989; 171:177–187.

68. George B, Thurel C, Pierron D, et al. Frequency of primarybrainstem lesions after head injuries: a CT analysis from186 cases of severe head trauma. Acta Neurochir (Wien)1981; 59:35–43.

69. Caplan LR, Zervas NT. Survival with permanent midbraindysfunction after surgical treatment of traumatic subduralhematoma: the clinical picture of a Duret hemorrhage. AnnNeurol 1977; 1:587–589.

70. Freytag E. Autopsy findings in head injuries from bluntforces: statistical evaluation of 1,367 cases. Arch Pathol1963; 75:402–413.

33

CT for ThromboembolicDisease: Protocols,

Interpretation, and Pitfalls1

Pulmonary embolism (PE) is, unfortunately, a common event. Approximately 600.000episodes of PE occur each year in the United States, and it is considered the third mostcommon acute cardiovascular event, after cardiac ischemia and stroke. However, PE is amuch more complex diagnostic challenge than ischemic heart disease and stroke. Com-plicating its diagnosis is the fact that PE is really only one manifestation of a larger dis-ease entity, venous thromboembolism, which also includes deep venous thrombosis.Until recently, the diagnosis of PE was made predominantly with pulmonary arteriogra-phy or ventilation-perfusion scintigraphy (ventilation-perfusion scans). Important limi-tations to each of these techniques precluded diagnosis of PE in many patients. The di-agnosis of deep venous thrombosis was approached separately, and many techniqueshave been used in its diagnosis, including physical examination, impedance plethys-mography, contrast venography, and ultrasonography (US). Each of these techniqueshas its own limitations.

Since the early 1990s, a new approach to the diagnosis of venous thromboembo-lism has developed, as computed tomographic (CT) pulmonary arteriography wasadded to the armamentarium of diagnostic techniques for PE. Multi–detector row CThas since increased the image quality and popularity of this technique. CT pulmonaryarteriography is an increasingly common examination, as everyone who cares for pa-tients in emergency settings knows. To improve the efficiency of diagnosis of thewhole spectrum of venous thromboembolism, a second CT technique has also beendeveloped: delayed CT through the deep venous system of the pelvis and lower ex-tremities, or “indirect CT venography.”

CT pulmonary arteriography and CT venography are, in some ways, straightforwardtechniques, and they appeal to clinicians because the direct visualization of clot withina vessel is immediately comprehensible. However, there are some challenges involvedwith using and refining these techniques. It is important to be familiar with data con-cerning sensitivity and specificity of CT pulmonary arteriography. These data help toshape an imaging algorithm for patients suspected of having venous thromboembo-lism, placing CT pulmonary arteriography in an appropriate context. An approach toreviewing images must be selected—an increasing challenge given the large data setsnow being produced with multi–detector row CT scanners. It is important for readersto be aware of the findings of acute and chronic thromboembolic disease and to beaware of artifacts and other findings that may simulate clots, both in the pulmonaryvessels and in the lower extremities. Distinctive features of chronic PE and deep

Lacey Washington, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 33–45.

1From the Department of Radiology, Medical College of Wisconsin, 9200 W Wisconsin Ave, Milwaukee, WI 53226-3596(e-mail: lwash@mcw.edu).

Was

hin

gto

n

34

venous thrombosis must also be recognized, so that pa-tients do not receive excessive or unnecessary treatment.

SENSITIVITY AND SPECIFICITY OF CTTECHNIQUES

CT Pulmonary Arteriography

CT evaluation for PE has become popular rapidlyfor a number of reasons: CT is readily available, theimages are easily understood, and alternate diagnosesare commonly made at CT in patients without evidentthromboembolic disease (1,2). Some authors, how-ever, continue to argue that the popularity of CT pul-monary arteriography is not well justified by data.There are two substantial reasons for the difficulty inproving the validity of CT techniques: One is the lackof a robust reference standard; the second is the rapidevolution of CT technology.

The only true reference standard for the diagnosis ofPE is autopsy. For obvious reasons, it is not possibleto validate CT with autopsy in any large series of hu-man patients. Autopsy data have other less obviouslimitations. The rate of detection of PE with autopsydepends to a large extent on the amount of attentiongiven to small vessels. If careful attention is given tothe smallest vessels, as many as 90% of patients mayhave emboli, either acute or chronic (3). The lung isthought to act normally as a filter, preventing embolifrom reaching the systemic arterial circulation, and itis unlikely that emboli are contributing factors in 90%of all deaths. In any individual case, it is difficult toassess the contribution of an embolus to the death ofa patient. In addition, because this level of attentionis seldom given to the pulmonary vessels, the use ofautopsy to exclude PE as a contributing factor in thedeath of any given patient is problematic—smallemboli may be present but not found.

In most studies of CT pulmonary arteriography, in-vestigators have compared CT pulmonary arterio-graphic results with pulmonary angiography, consid-ering angiography to be the reference standard be-cause it is the existing validated imaging technique.Pulmonary angiography is accepted because it hasbeen shown to be safe to withhold anticoagulationtherapy in patients with negative pulmonary angiogra-phy. However, the results of studies in the 1990s havesuggested that arteriography has its own limitations.Interreader variability for arteriography is substantial(4,5). In at least one study, investigators have at-tempted to approach the problem of anatomic truth.In a study with an animal model, a methacrylate castof pulmonary vessels was analyzed after the animalswere sacrificed, and both CT and angiographic resultswere compared with the autopsy results. CT was assensitive as angiography and had a comparable posi-tive predictive value (6). Interestingly, if angiographywas assumed to be the reference standard, the appar-

ent sensitivity of CT dropped to values similar tothose reported in clinical trials in humans that use ar-teriography as a reference standard.

The second problem with assessing CT has been therapid evolution of technology. In multiple studies, in-vestigators have compared single-detector helical CTwith pulmonary angiography. Early studies of the sen-sitivity of CT with respect to angiography reportedsensitivities of single-detector helical CT for PE in therange of 90%, when investigators were predominantlylooking at segmental or more central emboli (7–9).The results of occasional studies have shown lowersensitivities, usually because of a decreased sensitivityfor small subsegmental emboli (10). In relatively fewstudies have investigators even attempted to assess theaccuracy of newer multi–detector row CT scanners (11–13). Increasingly short scan times with multi–detectorrow CT scanners allow thin-collimation images to beobtained through the entire volume of the lungs. Im-ages with thinner collimation result in higher detectionrates of PE and in improved interreader agreement(14). As expected, two–detector row CT is more accu-rate than single–detector row CT when each is com-pared with angiography (12), and the results of stud-ies and subjective opinion suggest that increasing thenumber of CT detector rows, with the correspondingdecrease in scan times, will further increase accuracy.

Ironically, while research is attempting to determinewhether multi–detector row CT can adequately detectsmall subsegmental emboli, actual practice seems to befaced with the opposite problem: how to care for pa-tients with small, isolated subsegmental pulmonaryemboli found at multi–detector row CT (Fig 1). Theoverall clinical importance of subsegmental emboli iscontroversial, and the prevalence of isolated subseg-mental emboli differs in various studies, probably re-flecting differences among study populations. In stud-ies that used pulmonary angiography, isolated subseg-mental PE appears to occur in less than 10% of patients(4,12,15). Some of the emboli now found with multi–detector row CT clearly would have been undetectedwith previous imaging modalities, particularly in pa-tients with extensive lung disease. Because such patientswould have had other possible explanations for theirsymptoms, clinicians would have been reluctant tosubject such patients to angiography, and scintigraphyis unlikely to yield diagnostic results in patients withsubstantial lung disease. As a result, there are no goodhistorical data about the importance of these emboli.

Patients with good cardiopulmonary reserve and in-determinate ventilation-perfusion scans who have noevidence of deep venous thrombosis on serial non-invasive studies of the lower extremities have clini-cally good outcomes (16). Certainly, some of thesepatients can be assumed to have PE. It is thereforeclear that in the absence of deep venous thrombosis,patients with good cardiopulmonary reserve tolerate

CT fo

r Thro

mb

oem

bo

lic Disease

35

subsegmental emboli. Unfortunately, this result doesnot apply to patients with poor cardiopulmonary re-serve, even in the absence of deep venous thrombosis.It also does not apply to patients with a single nega-tive lower extremity US study and cannot be readilyextrapolated to CT venographic results.

At this time, the Prospective Investigation of Pulmo-nary Embolism Diagnosis (PIOPED) II trial is underway in an attempt to meet the need for large-scale pro-spective studies of CT pulmonary arteriography, withpreliminary results soon to be presented. In addition,the results of numerous outcome studies have shownlow rates of morbidity and mortality from subsequentPE in patients with no evidence of PE at CT pulmo-nary arteriography who did not receive anticoagulanttherapy (17–24). Because of doubts about the accu-racy of angiography, these outcome studies may even-tually become more important than those that use an-giography as a reference standard.

CT Venography

Until the late 1990s, deep venous thrombosis was anoccasional incidental finding at abdominal and pelvicCT examinations, and “direct” CT venography with in-jections into the pedal veins had been investigated asan alternative to conventional contrast venography andto US (25). However, in 1998, Loud et al (26) intro-duced a technique that took advantage of the contrastmaterial administered for CT pulmonary arteriographyto assess the lower extremity veins. With the introduc-tion of this technique (called “indirect” CT venogra-phy), CT became a single practical, clinically availabletest for both deep venous thrombosis and PE.

Because CT venography is a newer technique thanCT pulmonary arteriography, there are fewer studiesassessing its accuracy. However, the findings fromthose studies show moderately good interobserveragreement (27), with good sensitivity and specificity,when compared with US (28–32). An important ad-

vantage of CT venography over US is that the pelvicveins and profunda femoris veins are also imaged. Thefindings from some magnetic resonance (MR) veno-graphic studies have shown a higher than previouslysuspected incidence of isolated pelvic vein thrombo-sis, suggesting that imaging these veins may have asubstantial effect on patient care (33,34).

DIAGNOSTIC ALGORITHM

At the author’s institution, the following algorithm isused in evaluating patients suspected of having throm-boembolic disease. First, those patients who presentwith signs and symptoms of deep venous thrombosisalone (with no clinical evidence of PE) undergo USimaging because of the high sensitivity of lower ex-tremity US in this patient population.

Patients who present with signs and symptoms sug-gestive of PE initially undergo chest radiography.Those with normal chest radiographs are referred forscintigraphy. It has been shown that patients with anormal chest radiograph are more likely to have a de-finitive ventilation-perfusion scan result (ie, normalfindings, very low probability of PE, or high probabil-ity of PE) (35,36). Patients with abnormal chest radio-graphs undergo CT scans that include evaluation ofthe pulmonary arteries and the lower extremity veinsunless there is a contraindication to the intravenousadministration of contrast material, in which case pa-tients may be referred for US and/or scintigraphy asthe initial evaluation.

Patients who have nondiagnostic scintigraphy or CTmay be referred for further testing as indicated. For ex-ample, if enhancement of the lower extremity veins isless than optimal and the clinical scenario is suggestiveof the possibility of deep venous thrombosis, or if thereis substantial mixing artifact on the CT scan and thepresence of deep venous thrombosis is questioned, thepatient is referred for US. Although scintigraphy has

Figure 1. Isolated subseg-mental acute PE. (a, b) Axial CTimages show complete filling de-fect in subsegmental right lowerlobe pulmonary artery (arrow)that extends for several contigu-ous images.

Was

hin

gto

n

36

not been shown generally to provide additional infor-mation in patients who have undergone CT, in specificcases in which a single area is questioned at CT, scintig-raphy may be helpful in evaluating perfusion to a par-ticular portion of the lungs. Patients may undergo pul-monary arteriography in cases in which poor pulmo-nary arterial enhancement or severe motion artifactsrender CT pulmonary arteriography nondiagnostic. Oc-casionally, limited arteriographic studies are suggestedfor evaluation of a single vessel in which an embolus isquestioned at CT pulmonary arteriography.

When CT pulmonary arteriography and CT venogra-phy are of less than optimal quality, recent data fromour institution show that clinicians tend to behave asif PE is confidently excluded (37). In this small retro-spective series, this behavior did not have any negativeconsequences. Nevertheless, this approach is certainlynot well validated.

Although studies are continuing to demonstrategood outcomes in patients who do not receive anti-coagulant therapy after negative findings at CT pul-monary arteriography, some published diagnostic al-gorithms continue to allow for the potential risks ofmissed PE at CT. To account for potential false-negativestudies, patients may be stratified according to thepretest probability of PE (38). According to one suchstrategy, patients with a high pretest probability andnormal ventilation-perfusion scan results can besafely assumed to have no PE. In contrast, patientswith a high or intermediate pretest probability andnegative findings at CT are grouped with those whohave low- or intermediate-probability ventilation-per-fusion results and undergo lower extremity US. Unlessdeep venous thrombosis is demonstrated, this algo-rithm requires pulmonary angiography in this groupof patients. Patients with a low pretest probability ofPE may undergo a highly sensitive D-dimer assay—which is considered sufficiently sensitive to excludevenous thromboembolism only in patients with a lowpretest probability of PE. Patients with positiveD-dimer assays or those seen at institutions where thesensitive version of such an assay is not available un-dergo CT pulmonary arteriography or ventilation-per-fusion scanning. Patients with negative findings at CTpulmonary arteriography undergo lower extremity US,but in contrast to those patients with a high pretestprobability of venous thromboembolism, in the low-probability group the combination of negative find-ings at CT pulmonary arteriography and negative find-ings at lower extremity US is considered to excludevenous thromboembolism.

As more data accumulate, it is likely that fewer algo-rithms will insist on pulmonary angiography in pa-tients with negative findings on good-quality multi–detector row CT examinations. Recent guidelines fromthe British Thoracic Society (39) state that “patientswith a good quality negative CTPA do not require fur-

ther investigation or treatment for PE.” A more conser-vative conclusion is drawn by Kearon (40), who con-cludes that on the basis of results with single-detectorCT pulmonary arteriography, negative findings at CTand at lower extremity US in combination exclude PE,but only in patients with a low or intermediate pretestprobability of venous thromboembolism.

CT venography, as a newer technique, is not in-cluded in most published algorithms for the diagnosisof venous thromboembolism. At our institution, never-theless, it is routinely performed. This approach is con-troversial because the results of some studies have sug-gested that there is relatively little diagnostic yield forroutine CT venography. A recent review of 1435 patientsscanned during a 27-month period at our institutionshowed that 51 patients (3.6%) had deep venousthrombosis at CT venography but no PE at CT pulmo-nary arteriography (37). A separate study of 609 patientsat the same institution showed that 11% (21 patients)of the 183 patients with venous thromboembolism atCT had deep venous thrombosis alone, with no PE(41). While this finding is probably, in itself, an ad-equate rationale for the use of CT venography, the tech-nique becomes even more useful if clinicians are follow-ing an algorithm that assumes poor negative predictivevalue of negative findings at CT. Because these imagingalgorithms require evaluation for deep venous throm-bosis in all patients with negative findings at CT pulmo-nary arteriography, the use of CT venography will pre-vent a large number of lower extremity US examina-tions—at the cost of some additional radiation to thepatient. The amount of additional radiation will de-pend on the design of the CT venographic protocol, andthe effect of the additional radiation can be limited byavoiding CT venography in young patients and in pa-tients who have already undergone lower extremity US.

CT PULMONARY ARTERIOGRAPHIC TECHNIQUE

The rapid changes in CT scanner technology make itdifficult to comment on technical aspects of protocoldesign for CT pulmonary arteriography in a chaptersuch as this one, because there are so many possibledetector configurations. The portion of the lungs thatcan be imaged in a reasonable breath hold dependson the speed of the CT scanner. The earliest protocolswere designed for single-detector helical scanners,which only allowed imaging from the inferior pulmo-nary veins to the aortic arch in a reasonable time in-terval. As scanners improved, protocols were designedto include larger and larger volumes of the lungs, andmost recently, the entire volume of the lungs can beincluded in a single thin-section helical series with thenewest multidetector row CT scanners.

At the most basic level, protocols should be de-signed to image as much of the lung as is possible in areasonable breath-hold interval, with adequate peak

CT fo

r Thro

mb

oem

bo

lic Disease

37

kilovoltage and mAs to prevent excessive image noise.In general, thinner-collimation CT yields better depic-tion of peripheral vessels and greater sensitivity forsmall subsegmental PE. With multi–detector row CTscanners, subsecond images can be obtained with 1- or1.25-mm collimation, depending on the manufacturer,without excessive image noise, except in large patients.This results in better depiction of subsegmental andhigher-order vessels (42) and diminishes motion arti-facts (43). In larger patients, images with thin collima-tion may have unacceptable amounts of image noise.With multi–detector row CT scanners, the images canbe retrospectively combined to create thicker sectionsif there is excessive noise on initially viewed thin-sec-tion images. As CT tables become more tolerant of in-creased patient weight, there are more patients inwhom image noise is an important factor limiting thequality of CT imaging.

Dose can be decreased in smaller patients to limitradiation exposure. Beyond these specifics, there aremany parameters of CT pulmonary arteriographicprotocols that can be applied regardless of differencesamong scanners.

Scanning Parameters

Although not all investigators agree, some have sug-gested that caudal-to-cranial scanning improves thequality of many studies. Image degradation caused byrespiratory motion is usually most severe at the lungbases and least severe at the apices. With relatively slowscanner speeds that necessitate long breath-hold inter-vals, it is theoretically desirable to image the lower por-tion of the lung early in the breath hold, to minimizemotion artifact if the patient cannot maintain a breathhold for the entire imaging time. The use of oxygentherapy administered with nasal cannula may help pa-tients maintain longer breath holds, thus reducing mo-tion artifact. With the newest most rapid scanners, thelength of the breath hold has become important onlyin patients with the worst dyspnea. In this group, scan-ning during quiet breathing may be more successfulthan a failed attempt at breath holding.

Theoretically, patients undergoing mechanical venti-lation can be held in apnea with chemical paralysis. Ifthis is done, respiration should ideally be suspended athigh lung volumes, to increase pulmonary resistanceand improve opacification (44). At our institution, me-chanical ventilation is almost never suspended. Instead,mechanical ventilators are set to minimal tidal volumeand rate for the duration of the scan.

Timing of Imaging

At many institutions, fixed scan delays have beenfound to be adequate, with only a minority of patientshaving poorly enhanced pulmonary arteries if imag-ing is obtained at a delay of, for example, 28 secondsfrom the start of contrast material administration

(45). However, with four– or eight–detector row CTscanners, we believe that image quality seems to beimproved by the use of scan delays tailored to indi-vidual patients. There are two approaches to timingcontrast administration: (a) a preliminary time-den-sity curve or (b) commercially available bolus-trackingsoftware. With both approaches, initial nonenhancedimages are obtained to locate the pulmonary artery.To create a time-density curve, 10 low-dose images arethen obtained over the main pulmonary artery duringthe injection of 18 mL of contrast material. The timedelay for the diagnostic study is calculated by using atime to peak enhancement +5 seconds. With bolus-tracking software, contrast material is injected, andlow-dose images are obtained over the pulmonary ar-tery until contrast material appears, at which time di-agnostic imaging is initiated. The difficulty with theuse of bolus-tracking software is the unpredictable na-ture of the start of breath holding, which may lead torelatively little advance warning to the patient aboutthe start of a breath hold and thus to respiratory mo-tion on the initial images.

If a fixed delay is used, it is helpful to increase thelength of the delay or to obtain a time-density curvein patients who are thought to have cardiac dysfunc-tion, pulmonary arterial hypertension, or centralvenous stenoses. The short scan times used with therapid, new multi–detector row CT scanners have de-creased the need for precise timing. With an eight–detector row CT scanner, a volume through the lungscan be scanned in approximately 10–12 seconds, andwith a 16–detector row scanner, the lungs can bescanned in 5–6 seconds. If 120 mL of intravenous con-trast material is injected at a rate of 4 mL/sec, it is pos-sible to initiate scanning at a delay of 25 seconds, alonger delay than was almost ever necessary with care-fully timed contrast material boluses, and to finishscans within a few seconds of the end of the injection,essentially guaranteeing that imaging will not be per-formed too late in contrast material administration.

Contrast Material

Nonionic intravenous contrast material is used al-most always for CT pulmonary arteriography. Early inthe experience with CT pulmonary arteriography, someauthors recommended using low concentrations ofcontrast material at high flow rates to reduce streak arti-fact from high contrast material density in the superiorvena cava (46); however, this artifact is seldom nowconsidered to be particularly problematic. At our insti-tution, 120 mL of nonionic contrast material (iohexol;360 mg/mL) is injected at a rate of 4 mL/sec. Patientswith a mildly elevated serum creatinine are given 125mL of iso-osmolar contrast material (iodixanol; 320mg/mL), with hydration both before and after the ex-amination. Protocols ranging from injection rates of 2mL/sec to 5 mL/sec have been advocated by various

Was

hin

gto

n

38

investigators (45). With faster scanners, smaller quanti-ties of intravenous contrast material may adequatelyopacify the pulmonary arteries. However, at least 100mL and probably 120 mL must be given to achieve ad-equate enhancement of the lower extremity veins at CTvenography (and in larger patients, even this quantity issometimes inadequate). The average CT attenuation ofthrombus has been reported to range from 31 to 50 HU(30,47), and Bruce et al (48) have reported mean en-hancement values of between 91 and 97 HU with ad-ministration of 150 mL of iodinated contrast material,which should allow for the visualization of thrombus.

CT VENOGRAPHY

Technique

As with CT pulmonary arteriography, multiple dif-ferent protocols have been used by different investiga-tors to image the lower extremity veins. Some investi-gators advocate continuous helical imaging from thelevel of the renal veins to the level of the popliteal fos-sae (31). However, Loud and colleagues (26), whoperformed the first studies of CT venography, ob-tained only noncontiguous axial 10-mm images at 5-cm intervals. In our experience, continuous helical im-aging through the pelvis creates partial-volume arti-facts secondary to the obliquity of the iliac veins, andthis makes these images more difficult to interpret.For this reason, a compromise protocol has beenimplemented: 5-mm-thick axial CT images at 2-cm in-tervals are obtained from the level of the iliac crests to

the level of the popliteal fossae. When the veins areenhanced to 80 HU, contrast is considered to be ad-equate to exclude thrombosis; at lower degrees of en-hancement, clots may be diagnosed on occasion, butthey cannot be confidently excluded.

Initial studies of indirect CT venography were per-formed with imaging at 3 minutes after the initiationof contrast material administration (26). Some au-thors have advocated earlier imaging (49,50), whileothers have suggested that a delay of 4 minutes maybe preferable, at least in patients suspected of havingabnormal hemodynamics or slow flow (32). Timingfor CT venography appears, however, to be much lesscrucial than that for CT pulmonary arteriography.

Image Viewing

The large numbers of images generated with CT pul-monary arteriography and CT venography are most eas-ily evaluated on a workstation. At institutions that usepicture archiving and communications systems (PACS)for all clinical work, CT pulmonary arteriographic im-ages, like images from other studies, are stored andcalled back up from the PACS system. Workstation re-view is essential for accurate interpretation of CT pul-monary arteriographic studies. Even prior to the instal-lation of a PACS system, the practice at our institutionwas to review all studies on a CT workstation. Althoughselected images from the studies (every third image) arestill filmed at both lung and mediastinal windows, it isimpractical to interpret CT pulmonary arteriographicstudies on hard-copy images. Even when other cases

Figure 2. Paddle-wheel reconstruction. (a) Sagittal reformation is used as scout. Slab reformations are obtained rotating aroundcentral axis. (b) On slab image, webs (black arrows), narrowed artery (white arrows), and areas of focal stenosis (arrowheads) areseen in this patient with chronic thromboembolic pulmonary hypertension.

CT fo

r Thro

mb

oem

bo

lic Disease

39

were being reviewed or read on hard-copy images, CTpulmonary arteriographic studies were always restoredfrom optical disks, and current and comparison studieswere reviewed together on the workstation. If onlysmall clots are found, selected magnified demonstra-tion images may be filmed as a record. The theoreticalutility of these hard-copy images is that the images canbe taken with the patient to another facility; however,increasing numbers of institutions are supplying pa-tients and other facilities with images on CDs ratherthan on film, and CDs will probably continue to sup-plant the frequently awkward film record.

Postprocessing Techniques

Multiplanar reformations may occasionally be usedto clarify diagnoses at CT pulmonary arteriography.These reformations are most commonly used when itis difficult to separate perihilar lymphatic soft tissuefrom mural thrombus or wall thickening, particularlyin patients who are suspected of having chronic PE.Multiplanar reformations also can be useful in sepa-rating motion artifacts from real emboli. One grouphas advocated use of a “paddle-wheel” reformationtechnique, to try to depict all of the vessels in the

plane of section (51) (Fig 2). Most radiologists are ac-customed to reading axial images and are reluctant toattempt to use other imaging planes for diagnosticevaluation of images. However, CT pulmonary arteri-ography with multi–detector row CT scanners pro-duces such large numbers of images that there willprobably be increasing interest in alternate ways ofviewing these images.

CT FINDINGS IN PE

Acute PE

The findings of PE at CT are the same as those de-scribed in the earliest discussions of PE (46), althoughwith current technology, these findings may be soughtin much smaller vessels. Acute pulmonary embolimay cause partial or complete filling defects; they alsomay create a “railway-track sign” or mural defects.

A partial filling defect is a clot seen in the center of avessel surrounded by contrast material (Fig 3). Whenthe entire artery fails to opacify because of a centralclot, this is a complete filling defect. In the setting ofacute PE, arteries that are completely filled with clotmay be distended and appear larger than vessels of thesame generation that are free of emboli (Fig 4). A rail-way-track sign is a clot floating within a vessel, sur-rounded by contrast material, and is commonly seenin central vessels that are parallel to, rather than per-pendicular to, the plane of section. Mural defects areclots that adhere to the wall of the vessel. Mural em-boli are frequently seen in chronic as well as in acutePE; acute mural emboli commonly make acute angleswith the vessel walls. Emboli with several of thesemanifestations are seen in Figure 5.

Secondary signs of acute PE include depiction of theCT equivalent of a Hampton hump, the pulmonary“infarct.” These peripheral wedge-shaped areas of con-solidation represent hemorrhage, frequently withouttrue tissue necrosis, and are seen in only a minority ofpatients. Nonspecific findings in patients with PE in-clude atelectasis and pleural effusions; these findingsare common in patients with and without PE.

Figure 3. AcutePE. Axial CT im-age shows bilateralupper lobe partialfilling defects (ar-rows). Note also asmall complete fill-ing defect in sub-segmental right up-per lobe vessel.

Figure 4. Sub-segmental acutePE. Axial CT im-age shows thatsubsegmental rightupper lobe artery(arrow) is com-pletely filled withclot and appearsdistended. Re-maining vessels atsame level (arrow-heads) are well en-hanced and small-er in diameter.

Figure 5. Mul-tiple bilateral acutepulmonary emboli.Axial CT imageshows incompletefilling defect(straight arrow) inright basilar arterytrunk. Mural throm-bus (arrowhead)forms acute angleswith vessel wall inleft anteromedialbasal segmentalartery. Completefilling defect(curved arrow) isseen in right me-dial basal segmental artery. Smaller emboli are seen in othervessels.

Was

hin

gto

n

40

The main differential diagnostic considerationswhen filling defects are seen in pulmonary arteries arechronic PE and artifact. In an emergency setting, itmay be valuable to distinguish small chronic emboli,which may not require anticoagulant therapy, fromsimilar acute emboli. In addition, patients may pre-sent with chronic thromboembolic pulmonary hyper-tension without any previously diagnosed acute PE,and it is important to separate these patients from pa-tients with massive acute PE, who may, perhaps, becandidates for thrombolytic therapy. Rarely, filling de-fects may represent neoplasms—either primary sarco-mas or central tumor emboli from neoplasms such ashepatocellular carcinoma or renal cell carcinoma.

Chronic PE

The most specific finding for chronic PE is calcifica-tion within a clot. However, this finding is insensitive,and unless there is considerable noncalcified muralthrombus, calcification may be difficult to detect inthe presence of contrast material enhancement (Fig 6).Other findings (and the clinical history) may be morehelpful in making this diagnosis.

Chronic emboli are frequently eccentric and contigu-ous with the vessel wall, and when they are eccentric,they more commonly make obtuse angles, rather than

Figure 6. Chronic PE. (a) Oncontrast-enhanced axial CT im-age, mural calcification in rightpulmonary artery is subtle.(b) Nonenhanced axial CT im-age shows mural calcification(arrows) clearly.

Figure 7.Chronic PE. AxialCT image showsbilateral centralmural emboli (ar-rows) making ob-tuse angles withthe vessel walls,suggesting chro-nicity.

Figure 8. Bilat-eral webs inchronic PE. AxialCT image showsthin linear fillingdefects (arrows)bilaterally in a pa-tient with chronicthromboembolicpulmonary hyper-tension. On asingle image,these filling de-fects may be mis-taken for vascular bifurcations, particularly when vessel has anirregular shape as a result of chronic PE. On contiguous axialimages, contrast material columns on both sides of web extendin parallel for several contiguous images.

Figure 9. ChronicPE. Axial CT imageshows that rightbasilar segmentalarteries (arrows)are completelynonenhanced andsmall in caliber. Incontrast, left basilarsegmental arteries(arrowheads) arenormal in caliber,although there issome mural thick-ening.

acute angles, with the vessel wall (Fig 7). Vascular websare thin linear or planar filling defects and strongly sug-gest chronicity (Fig 8). Areas of arterial stenosis may beevident. Stenosis can be difficult to perceive on imagesobtained perpendicular to a vessel plane but should besuspected when an enhancing vessel is markedly smallerthan the adjacent bronchus and other vessels of a simi-lar generation. Multiplanar reformations may help to

CT fo

r Thro

mb

oem

bo

lic Disease

41

image these. Stenosis in isolation is not diagnostic ofchronic PE but may support the diagnosis when otherfindings are present. Abrupt cutoff of the contrast col-umn in a vessel may be seen, with a narrowed (ratherthan dilated) nonenhancing distal vessel (Fig 9). Vesselscontaining chronic emboli are apt to be smaller thanuninvolved vessels of the same order, in contrast to theenlarged vessels frequently seen with acute PE.

Dilation of the main pulmonary arteries is seen inpulmonary arterial hypertension, including that causedby chronic PE. There may be prominent bronchial col-lateral arteries. Mosaic attenuation in the lungs is morecommonly seen in chronic PE than in acute emboli.This finding is not specific for chronic PE, and morepatients with it will have small airways disease thanchronic PE. Mosaic attenuation may also be seen withother causes of pulmonary arterial hypertension. Expi-ratory images have been used to distinguish small air-ways disease from diseases with vascular obstruction;mosaic perfusion secondary to small airways diseasewill show a dramatic increase in contrast betweenlow- and high-attenuation regions on expiratory im-ages. In a recent article, however, investigators suggestedthat this change in contrast (air trapping) may also beseen in acute or chronic thromboembolic disease (52).Small subpleural areas of scar, which are probably se-quelae of old areas of infarction, are a nonspecific find-ing that is also common in patients with chronic PE.

ARTIFACTS AND PITFALLS

Multiple artifacts that could be confused with embolihave been described. To help avoid confusion, it isbest to look for sharply demarcated areas of low at-tenuation in vessels and to diagnose emboli onlywhen these are seen on more than one sequential im-age, at least on vessels imaged in cross section. Acuteemboli are unlikely to be so small that they will beseen in cross section for a distance of only 1–3 mm.

Many areas of confusion can be easily avoided by re-viewing scans on a workstation.

Anatomic Pitfalls

At least a minimal understanding of pulmonary arte-rial anatomic structures is necessary to interpret CT pul-monary arteriographic studies. However, there aremany variants of pulmonary arterial anatomic struc-tures, and the naming of vessels is frequently confusing.It helps to remember that an artery is usually named ac-cording to the segmental bronchus that it accompanies.Pulmonary veins course independent of the bronchi.

If imaging is performed early in the administration ofthe contrast material bolus, unopacified pulmonaryveins can be confused with PE (Fig 10). This mistakecan easily be avoided by workstation review, which willallow the reader to trace vessels back either to their ori-gins at the pulmonary arteries or to their terminationsat the left atrium. It is also helpful to keep in mind thatthe lower lobe pulmonary arteries are peripheral to theaccompanying bronchi, whereas lower lobe veins arecentral to the bronchi. In the upper lobes, the arteriesare central to the corresponding bronchi.

Another misinterpretation that can be avoided withthe use of workstation review is mistaking mucoid im-paction of a bronchus for a pulmonary embolus. Mu-coid impaction of the bronchus causes central lowerattenuation with peripheral higher attenuation, whichon a single image can look much like a pulmonaryembolus. Following a structure on lung window set-tings to a section on which it appears aerated helps tomake this distinction (Fig 11). Hilar lymph nodescould also potentially be mistaken for emboli, mostcommonly for chronic emboli, rather than acute em-boli. Again, workstation review helps to avoid this in-terpretive pitfall because lymph nodes will not extendthrough multiple sections.

The most common imaging artifacts seen on CTpulmonary arteriographic studies are streak artifacts

Figure 10. Unopacified pulmo-nary veins. (a) Image at CT pul-monary angiographic windowsetting shows both adequatelyenhanced vessels (arrows) andpoorly enhanced vessels (arrow-heads). (b) Axial CT imageviewed at lung window settingsdemonstrates that the poorly en-hanced vessels (arrowheads)are independent of and medialto bronchi, indicating that ves-sels are veins.

Was

hin

gto

n

42

and motion artifacts. Streak artifacts often arise fromdense contrast opacification of the superior vena cava ormay arise from calcified nodes, metallic surgical clips,pacemakers, and other implanted devices. These artifactscan cause focal areas of low attenuation in the vessel.Streak artifacts are seldom sufficiently well defined totruly cause confusion with emboli and usually continuebeyond the vessel and do not extend through multipleconsecutive images.

Motion artifacts, on the other hand, can convincinglysimulate emboli. Partial volume averaging of the lungsurrounding vessels occurs with motion, causing low at-tenuation to appear to be located within an otherwiseenhanced vessel. The easiest way to avoid confusing mo-tion artifact with emboli is to evaluate scans on a work-station and switch window and level settings from thesoft-tissue windows used to evaluate for PE to lung win-dows. Motion artifacts are much more easily seen atlung window settings. In addition, workstation reviewwill enable the reader to see that the vessel changes posi-tion rapidly from one section to another, confirmingthat there is motion. Multiplanar reformations may alsodemonstrate motion. The area most prone to motion ar-tifact is the portion of the left lung immediately behindthe heart that receives transmitted cardiac pulsations.

Other Pitfalls

It is important to avoid interpreting poor enhance-ment of vessels as emboli. This pitfall is usually easyto avoid because the degree of enhancement is gener-ally uniform throughout the images or may perhapsbe uniformly low in the earliest images or the latestimages, suggesting imaging either too early or too latewith respect to contrast enhancement. Higher levels ofpulmonary vascular resistance lead to better enhance-ment, and imaging at maximal inspiration is recom-mended to increase pulmonary vascular resistance andoptimize arterial enhancement.

There may, however, be a pitfall with deep inspira-tion. Recently, a new artifact has been described, whichis most commonly seen in young patients. This artifactis thought to be caused by rapid, deep inspiration (of-ten occurring at the beginning of a breath hold), whichresults in negative intrathoracic pressures and an influxof unopacified blood from the inferior vena cava, dilut-ing the contrast material bolus (53). This is particularlyconfusing because the vessels will be well enhancedmore proximally, and there may be good pulmonaryvenous enhancement, indicating that the delay to theinitiation of imaging is adequate. This artifact shouldbe considered as a possible cause when multiple poorlyenhanced arteries are seen at the same anatomic level,and the artifact can theoretically be prevented by care-ful instructions to patients to avoid a rapid deep inspi-ration at the beginning of a scan.

Another potential cause of confusion is the partialvolume averaging that occurs at vascular bifurcationsor in small vessels seen in the axial plane. A low-at-tenuation region at a vascular bifurcation that doesnot extend to more than one or two images shouldnot be interpreted as PE. Multiplanar reformationsmay be helpful in distinguishing these low-attenua-tion regions from true emboli.

Intravascular Neoplasm

Intravascular tumor is a rare cause of intraarterial fill-ing defects. Occasionally, large central intraarterial me-tastases will occur; this is most common in hepatocellu-lar and renal cell carcinomas. The possibility of intra-arterial metastasis should be considered in patients withother evidence of metastatic disease from one of thesetwo primary tumors. Primary sarcomas of the pulmo-nary arteries are rare. Marked expansion of the vesseland large central unifocal filling defects may suggest thepossibility of pulmonary artery sarcoma. When intra-arterial neoplasm is suspected, delayed imaging

Figure 11. Mucoid impaction of bronchi. (a) Axial CT image at soft-tissue window setting shows low-attenuation tubular structures(arrows) adjacent to enhanced pulmonary arteries. (b, c) Axial CT images viewed at lung window settings show low-attenuationstructures (arrows) continuing to aerated regions, indicating that the structures represent mucus-filled bronchi. They are also seen torun parallel to arteries on contiguous images.

CT fo

r Thro

mb

oem

bo

lic Disease

43

through the pulmonary arteries may show enhance-ment of the low-attenuation region, indicating that itis not bland but tumor thrombus.

FINDINGS OF DEEP VENOUS THROMBOSIS

Like PE, deep venous thrombosis appears at CT as afilling defect in an otherwise enhanced vessel. A fillingdefect may be focal or partial, or the vein may be com-pletely filled with clot. Like an artery, a vein that iscompletely filled with acute thrombus is usually en-larged. Findings of acute deep venous thrombosis thatare not seen in PE include perivenous stranding andmural enhancement (Fig 12). Stranding in the fataround the vein is caused by perivenous edema asso-ciated with acute deep venous thrombosis. These find-ings should be sought not only in the portions of thevenous system easily accessible to compression US butalso in the pelvis and in the profunda femoris veins.

Like chronic PE, chronic deep venous thrombosismay result in calcification of the vein. There may beprominent collateral veins. The vein is usually smallerthan the accompanying artery, in contrast to the nor-mal state and to the distension seen with acute deepvenous thrombosis. If the vein appears markedly lesswell enhanced than the artery, a Hounsfield unit mea-surement should be obtained; imaging may have oc-curred early in the circulation of contrast material,and caution should be used in interpreting the study.Unfortunately, similar findings may be seen in pa-tients with extensive bilateral thrombosis; a few de-layed images may help to distinguish artifact from realthrombosis, and examination of the entire venous sys-tem for areas of better enhancement may also be help-ful. Artifacts that may be seen at CT venography in-clude mixing and streak artifacts. Streak artifacts maybe particularly problematic in patients with orthope-

dic hardware; unfortunately, this patient population isat increased risk for deep venous thrombosis. Thesepatients may require additional imaging with US ifsuch imaging is clinically feasible.

In conclusion, clinical practice with respect to evalu-ation of patients with venous thromboembolism ischanging rapidly, and CT pulmonary arteriographyand CT venography are increasingly important partsof radiologic practice in emergency settings. As morestudies validate the safety of withholding anticoagu-lant therapy from patients with negative CT findingsfor PE, CT pulmonary arteriography will probablycontinue to increase in popularity. CT of the lower ex-tremity veins may add to the confidence with whichvenous thromboembolism is excluded in a given pa-tient. Using multi–detector row CT scanners increasesdepiction of small peripheral vessels, and workstationreview increases confidence in the diagnosis or exclu-sion of PE. Familiarity with the findings of venousthromboembolism and with potential diagnostic pit-falls will ensure accurate interpretation of these evermore common studies.

References1. van Rossum AB, Pattynama PM, Mallens WM, Hermans J,

Heijerman HG. Can helical CT replace scintigraphy in thediagnostic process in suspected pulmonary embolism? aretrospective-prospective cohort study focusing on total di-agnostic yield. Eur Radiol 1998; 8:90–96.

2. Cross JJ, Kemp PM, Walsh CG, Flower CD, Dixon AK. Arandomized trial of spiral CT and ventilation perfusion scin-tigraphy for the diagnosis of pulmonary embolism. ClinRadiol 1998; 53:177–182.

3. Wagenvoort CA, Mooi WJ. Biopsy pathology of the pulmo-nary vasculature. London, England: Chapman & Hall, 1989.

4. Diffin DC, Leyendecker JR, Johnson SP, Zucker RJ, GrebePJ. Effect of anatomic distribution of pulmonary emboli oninterobserver agreement in the interpretation of pulmonaryangiography. AJR Am J Roentgenol 1998; 171:1085–1089.

5. Stein PD, Henry JW, Gottschalk A. Reassessment of pul-monary angiography for the diagnosis of pulmonary embo-lism: relation of interpreter agreement to the order of the in-volved pulmonary arterial branch. Radiology 1999; 210:689–691.

6. Baile EM, King GG, Muller NL, et al. Spiral computed to-mography is comparable to angiography for the diagnosis ofpulmonary embolism. Am J Respir Crit Care Med 2000;161:1010–1015.

7. Remy-Jardin M, Remy J, Deschildre F, et al. Diagnosis ofpulmonary embolism with spiral CT: comparison with pul-monary angiography and scintigraphy. Radiology 1996;200:699–706.

8. Mayo JR, Remy-Jardin M, Muller NL, et al. Pulmonary em-bolism: prospective comparison of spiral CT with ventila-tion-perfusion scintigraphy. Radiology 1997; 205:447–452.

9. Garg K, Welsh CH, Feyerabend AJ, et al. Pulmonary embo-lism: diagnosis with spiral CT and ventilation-perfusion scan-ning—correlation with pulmonary angiographic results orclinical outcome. Radiology 1998; 208:201–208.

10. Goodman LR, Curtin JJ, Mewissen MW, et al. Detection ofpulmonary embolism in patients with unresolved clinical andscintigraphic diagnosis: helical CT versus angiography. AJRAm J Roentgenol 1995; 164:1369–1374.

Figure 12. Acute deep venous thrombosis of left profundafemoris vein. Axial CT image shows that left profunda femorisvein (white arrow) is distended, with mural enhancement andperivenous edema. Note normal-caliber, normally enhancingright profunda femoris vein (black arrow).

Was

hin

gto

n

44

11. Coche E, Verschuren F, Keyeux A, et al. Diagnosis of acutepulmonary embolism in outpatients: comparison of thin-colli-mation multi–detector row spiral CT and planar ventilation-perfusion scintigraphy. Radiology 2003; 229:757–765.

12. Qanadli SD, Hajjam ME, Mesurolle B, et al. Pulmonary em-bolism detection: prospective evaluation of dual-section heli-cal CT versus selective pulmonary arteriography in 157 pa-tients. Radiology 2000; 217:447–455.

13. Blachere H, Latrabe V, Montaudon M, et al. Pulmonary em-bolism revealed on helical CT angiography: comparison withventilation-perfusion radionuclide lung scanning. AJR Am JRoentgenol 2000; 174:1041–1047.

14. Schoepf UJ, Holzknecht N, Helmberger TK, et al. Subseg-mental pulmonary emboli: improved detection with thin-colli-mation multi–detector row spiral CT. Radiology 2002; 222:483–490.

15. Stein PD, Henry JW. Prevalence of acute pulmonary embo-lism in central and subsegmental pulmonary arteries and re-lation to probability interpretation of ventilation/perfusionlung scans. Chest 1997; 111:1246–1248.

16. Hull RD, Raskob GE, Coates G, Panjuu AA, Gill GJ. A newnoninvasive management strategy for patients with sus-pected pulmonary embolism. Arch Intern Med 1989; 149:2549–2555.

17. Garg K, Sieler H, Welsh CH, Johnston RJ, Russ PD. Clinicalvalidity of helical CT being interpreted as negative for pulmo-nary embolism: implications for patient treatment. AJR Am JRoentgenol 1999; 172:1627–1631.

18. Lomis NN, Yoon HC, Moran AG, Miller FJ. Clinical outcomesof patients after a negative spiral CT pulmonary arteriogramin the evaluation of acute pulmonary embolism. J Vasc IntervRadiol 1999; 10:707–712.

19. Goodman LR, Lipchik RJ, Kuzo RS, Liu Y, McAuliffe TL,O’Brien DJ. Subsequent pulmonary embolism: risk after anegative helical CT pulmonary angiogram—prospective com-parison with scintigraphy. Radiology 2000; 215:535–542.

20. Gottsater A, Berg A, Centergard J, Frennby B, Nirhov N,Nyman U. Clinically suspected pulmonary embolism: is itsafe to withhold anticoagulation after a negative spiral CT?Eur Radiol 2001; 11:65–72.

21. Musset D, Parent F, Meyer G, et al. Diagnostic strategy forpatients with suspected pulmonary embolism: a prospectivemulticentre outcome study. Lancet 2002; 360:1914–1920.

22. van Strijen MJ, de Monye W, Schiereck J, et al. Single-de-tector helical computed tomography as the primary diagnos-tic test in suspected pulmonary embolism: a multicenter clini-cal management study of 510 patients. Ann Intern Med2003; 138:307–314.

23. Swensen SJ, Sheedy PF II, Ryu JH, et al. Outcomes afterwithholding anticoagulation from patients with suspectedacute pulmonary embolism and negative computed tomo-graphic findings: a cohort study. Mayo Clin Proc 2002; 77:130–138.

24. Tillie-Leblond I, Mastora I, Radenne F, et al. Risk of pulmo-nary embolism after a negative spiral CT angiogram in pa-tients with pulmonary disease: 1-year clinical follow-upstudy. Radiology 2002; 223:461–467.

25. Baldt MM, Zontsich T, Stumpflen A, et al. Deep venousthrombosis of the lower extremity: efficacy of spiral CTvenography compared with conventional venography in di-agnosis. Radiology 1996; 200:423–428.

26. Loud PA, Grossman ZD, Klippenstein DL, Ray CE. Com-bined CT venography and pulmonary angiography: a new di-agnostic technique for suspected thromboembolic disease.AJR Am J Roentgenol 1998; 170:951–954.

27. Garg K, Kemp JL, Russ PD, Baron AE. Thromboembolic dis-ease: variability of interobserver agreement in the interpreta-tion of CT venography with CT pulmonary angiography. AJRAm J Roentgenol 2001; 176:1043–1047.

28. Loud PA, Katz DS, Bruce DA, Klippenstein DL, GrossmanZD. Deep venous thrombosis with suspected pulmonary em-bolism: detection with combined CT venography and pulmo-nary angiography. Radiology 2001; 219:498–502.

29. Coche EE, Hamoir XL, Hammer FD, Hainaut P, Goffette PP.Using dual-detector helical CT angiography to detect deepvenous thrombosis in patients with suspicion of pulmonaryembolism: diagnostic value and additional findings. AJR Am JRoentgenol 2001; 176:1035–1038.

30. Duwe KM, Shiau M, Budorick NE, Austin JH, Berkmen YM.Evaluation of the lower extremity veins in patients with sus-pected pulmonary embolism: a retrospective comparison ofhelical CT venography and sonography—2000 AmericanRoentgen Ray Society Executive Council Award I. AJR Am JRoentgenol 2000; 175:1525–1531.

31. Cham MD, Yankelevitz DF, Shaham D, et al. Deep venousthrombosis: detection by using indirect CT venography—thePulmonary Angiography-Indirect CT Venography CooperativeGroup. Radiology 2000; 216:744–751.

32. Garg K, Kemp JL, Wojcik D, et al. Thromboembolic disease:comparison of combined CT pulmonary angiography andvenography with bilateral leg sonography in 70 patients. AJRAm J Roentgenol 2000; 175:997–1001.

33. Spritzer CE, Arata MA, Freed KS. Isolated pelvic deepvenous thrombosis: relative frequency as detected with MRimaging. Radiology 2001; 219:521–525.

34. Stern JB, Abehsera M, Grenet D, et al. Detection of pelvic veinthrombosis by magnetic resonance angiography in patientswith acute pulmonary embolism and normal lower limb com-pression ultrasonography. Chest 2002; 122:115–121.

35. Lesser BA, Leeper KV Jr, Stein PD, et al. The diagnosis ofacute pulmonary embolism in patients with chronic obstruc-tive pulmonary disease. Chest 1992; 102:17–22.

36. Goldberg SN, Palmer EL, Scott JA, Fisher R. Pulmonary em-bolism: prediction of the usefulness of initial ventilation-perfu-sion scanning with chest radiographic findings. Radiology1994; 193:801–805.

37. Eyer B, Goodman LR, Washington L, Lipchik RJ. Isolatedsubsegmental pulmonary embolism (PE) or indeterminate PEdiscovered on helical CT: clinician response and patient out-come. AJR Am J Roentgenol (in press).

38. Fedullo PF, Tapson VF. Clinical practice: the evaluation ofsuspected pulmonary embolism. N Engl J Med 2003; 349:1247–1256.

39. British Thoracic Society Standards of Care Committee Pul-monary Embolism Guideline Development Group. BritishThoracic Society guidelines for the management of suspectedacute pulmonary embolism. Thorax 2003; 58:470–483.

40. Kearon C. Excluding pulmonary embolism with helical (spiral)computed tomography: evidence is catching up with enthusi-asm. CMAJ 2003; 168:1430–1431.

41. Cosmic MS, Goodman LR, Lipchik RJ, Washington L. Detec-tion of deep venous thrombosis with combined helical CTscan of the chest and CT venography in unselected cases ofsuspected pulmonary embolism. Presented at the 98th Inter-national Conference of the American Thoracic Society, At-lanta, Ga, May 17–22, 2002.

42. Ghaye B, Szapiro D, Mastora I, et al. Peripheral pulmonaryarteries: how far in the lung does multi-detector row spiral CTallow analysis? Radiology 2001; 219:629–636.

43. Boiselle PM, Rosen M, Raptopoulos V. Comparison of CTpulmonary angiography image quality on single and multi-detector scanners (abstr). In: American Roentgen Ray Soci-ety 101st Annual Meeting Abstract Book. Leesburg, Va:American Roentgen Ray Society, 2001; 20.

44. Remy-Jardin M, Remy J, Artaud D, Fribourg M, Beregi JP.Spiral CT of pulmonary embolism: diagnostic approach, inter-pretive pitfalls and current indications. Eur Radiol 1998;8:1376–1390.

CT fo

r Thro

mb

oem

bo

lic Disease

45

45. Yankelevitz DF, Shaham D, Shah A, Rademacker J,Henschke CI. Optimization of contrast delivery for pulmonaryCT angiography. Clin Imaging 1998; 22:398–403.

46. Remy-Jardin M, Remy J, Wattinne L, Giraud F. Central pul-monary thromboembolism: diagnosis with spiral volumetricCT with the single-breath-hold technique—comparison withpulmonary angiography. Radiology 1992; 185:381–387.

47. Loud PA, Katz DS, Klippenstein DL, Shah RD, GrossmanZD. Combined CT venography and pulmonary angiographyin suspected thromboembolic disease: diagnostic accuracyfor deep venous evaluation. AJR Am J Roentgenol 2000;174:61–65.

48. Bruce D, Loud PA, Klippenstein DL, Grossman ZD, Katz DS.Combined CT venography and pulmonary angiography: howmuch venous enhancement is routinely obtained? AJR Am JRoentgenol 2001; 176:1281–1285.

49. Matar LD, Ramirez JA, McAdams HP, Farrell MB, HerndonJE. Optimal timing of CT venography following CT pulmo-nary angiography using a multidetector row helical scanner:work in progress (abstr). Radiology 1999; 213(P):472.

50. Patel S, Kazerooni EA. Venous enhancement in the pelvisand thigh on multi-detector CT with variable scan delaytimes. In: Thoracic imaging 2000: proceedings of the 18thannual meeting of the Society of Thoracic Radiology, March12–16, 2000, San Diego, Calif. Rochester, Minn: Society ofThoracic Radiology, 2000; 73.

51. Chiang EE, Boiselle PM, Raptopoulos V, Reynolds KF,Rosen MP, Simon M. Detection of pulmonary embolism:comparison of paddlewheel and coronal CT reformations—initial experience. Radiology 2003; 228:577–582.

52. Arakawa H, Kurihara Y, Sasaka K, Nakajima Y, Webb WR.Air trapping on CT of patients with pulmonary embolism.AJR Am J Roentgenol 2002; 178:1201–1207.

53. Gosselin MV, Thieszen SL, Yoon CH. Contrast dynamicsduring pulmonary CT angiogram (PCTA): analysis of an in-spiration-induced artifact (abstr). In: Thoracic imaging 2001:proceedings of the 19th annual meeting of the Society ofThoracic Radiology, April 4–8, 2001, Boca Raton, Fla. Roch-ester, Minn: Society of Thoracic Radiology, 2001; 186.

NO

TES

47

CT of Nontraumatic AorticEmergencies1

Emergency evaluation of the aorta usually takes place in one of two clinical settings:the acute thoracic aortic syndrome or the suspected ruptured abdominal aortic aneu-rysm. Conventional angiography has been the reference standard in the past forboth clinical syndromes, but the past 2 decades have seen increasing roles for com-puted tomography (CT), magnetic resonance (MR) imaging, and ultrasonography(US). In the only modern study (to our knowledge) to compare modern techniquesfor detection of aortic dissection, investigators found that single–detector row helicalCT was equivalent to multiplanar transesophageal echocardiography and spin-echoMR imaging on a 0.5-T MR imager in its ability to detect aortic dissection (1). Allmodalities performed nearly perfectly, with 100% sensitivity for all modalities andhigh specificity, ranging from 94% to 100%. CT was superior to the other modalitiesin establishing the presence or absence of aortic arch involvement by dissection: Thesensitivity was 93% for CT, 60% for transesophageal echocardiography, and 67% forMR imaging; and specificity was 97% for CT, 85% for transesophageal echocardiog-raphy, and 88% for MR imaging (1). In the emergency center equipped with a modernCT scanner, CT is often the preferred modality for urgent evaluation of the aorta.

The use of CT in the nontraumatized emergency patient is the focus of this syllabuschapter. The chapter first reviews the pathology and imaging findings in nontraumaticemergency conditions of the aorta. The principles of helical single– and multi–detec-tor row CT evaluation of the aorta are then reviewed, and the chapter concludes withknown problems and pitfalls in the evaluation of the aorta with CT.

ACUTE AORTIC SYNDROME

Acute aortic syndrome is characterized by aortic pain coexisting with hypertension(2). Aortic pain is distinct from angina pectoris and is characterized by severe, in-tense, acute searing or tearing, throbbing, and migratory chest pain (3). Pain may ra-diate to the anterior portion of the chest, neck, throat, or jaw, particularly when thepain originates in the ascending aorta, and may radiate to the back and abdomenwhen the pain originates in the descending aorta. The pain associated with classicaortic dissection is similar to the pain experienced by patients with penetrating aor-tic ulcer and intramural aortic hematoma. Together, these conditions can be lumpedtogether as acute aortic syndrome, and imaging is usually performed for “chest pain;rule out aortic dissection.”

O. Clark West, MD, and Sanjeev Bhalla, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 47–57.

1From the Department of Radiology, the University of Texas Medical School at Houston, MSB 2.100, 6431 Fannin, Houston,TX 77030 (O.C.W.); and Mallinckrodt Institute of Radiology, Washington University School of Medicine, St Louis, Mo (S.B.).

Wes

t an

d B

hal

la

48

is much less common than extension of thoracic aor-tic dissection into the abdominal aorta (8).

Predisposing factors for aortic dissection includesystemic hypertension, bicuspid aortic valve, coarcta-tion of the aorta, Marfan syndrome, Ehlers-Danlossyndrome, Turner syndrome, giant cell arteritis, third-trimester pregnancy, cocaine abuse, trauma, intra-aortic catheterization, and previous aortic valve re-placement (4). Atherosclerosis is not a risk factor foracute aortic dissection.

Classification.—Both the Stanford and the DeBakeyclassification systems are widely used to describe aor-tic dissection (9,10). The Stanford system is simple,but the information conveyed in the DeBakey systemis important for complete description of the extent ofdissection.

Aortic dissections involving the ascending aorta areclassified as Stanford type A. Type A dissections mayalso involve the aortic arch and descending aorta (Fig1). Aortic dissections involving only the descendingaorta are classified as Stanford type B (Fig 2).

The more detailed DeBakey system effectively subdi-vides Stanford type A dissections into those involvingthe ascending aorta, aortic arch, and descending aorta(type I) and those confined to the ascending aorta (typeII). Stanford type B dissections are subdivided intothose that are confined to the descending thoracic aorta(type IIIa) and those that involve both the descendingthoracic aorta and the abdominal aorta (type IIIb). Ingeneral, untreated type A dissections have a high mor-tality rate and are managed with urgent surgical repair.Type B dissections are often managed with medicaltherapy, with elective surgical repair in selected cases.

Recognition of ascending aortic involvement (typeA vs type B) is of paramount importance in determin-ing initial management. The radiology report shouldalso include both the proximal and distal extent of thelesion. Dividing the aorta into zones allows clear re-porting. One practical system divides the aorta intothe aortic root, ascending aorta, aortic arch (from theorigin of the brachiocephalic trunk to the origin of theleft subclavian artery), proximal descending aorta(from the origin of the left subclavian artery to the topof the left atrium), distal descending aorta (from thetop of the left atrium to the diaphragm), juxtaceliac

Clinical diagnosis of aortic dissection is problem-atic. The results of a meta-analysis based on 21 ar-ticles published from 1996 to 2000 indicate thatmost patients with thoracic aortic dissection havesevere pain with an abrupt onset (4). The absence ofsudden onset decreases the probability that aorticdissection is present. Potentially useful symptomsof aortic dissection are (a) a “tearing” or “ripping”character of the pain or (b) pain that migrates fromone location in the chest to another. Physical find-ings are present in one-third or fewer of the cases(4). A pulse deficit or focal neurologic deficit indi-cates a high likelihood of thoracic aortic dissectionin the appropriate clinical setting. A normal aorticand mediastinal contour on the chest radiographdiminishes the likelihood that aortic dissection ispresent (4).

Aortic Dissection

Pathology.—The aorta and great vessels are com-posed of three layers of tissue: the intima, the media,and the adventitia. In acute aortic dissection, the lay-ers of the aortic wall are torn apart, creating a falselumen within the substance of the aortic media thatruns parallel to the true lumen (5). Aortic dissectionbegins with tearing in the aortic intima and the in-ner layer of the aortic media, which permits entry ofblood along a false lumen within the aortic media(5). The false lumen splits the aortic media into in-ner and outer layers. The intimomedial flap is com-posed of the intima and the attached inner layer ofthe media and separates the true and false lumina.The outer wall of the false lumen is composed of theouter layer of the media and the attached adventitia.

Intimal tears occur at points of high shear force, cre-ated by the jet of blood expelled under pressure fromthe heart (6). Most intimal tears occur in the ascendingaorta along the right lateral wall (7). The tear then pro-pagates along the greater curvature of the aortic archand continues caudally along the descending aorta.The tear may also propagate in a retrograde fashiontoward the aortic valve.

Tears originating beyond the ascending aorta mostoften occur immediately distal to the left subclavianartery (7). Isolated dissection of the abdominal aorta

Figure 1. Type A aortic dissection with intramural hematoma and cardiac tamponade. (a–d) Unenhanced transverse CT imagesshow (a) displaced intimal calcifications in aortic arch, (b) intramural hematoma in left half of wall of ascending aorta, and (c, d) largeamount of blood in pericardial sac. (e–h) Contrast-enhanced transverse CT images obtained during contrast material injection showintimomedial flap extending longitudinally along course of intimal calcification, separating anterior false lumen from posterior true lu-men. (f) False lumen on right side of ascending aorta is enhanced and compresses true lumen. (f–h) Posterior false lumen in de-scending aorta is unenhanced at this early stage of contrast material distribution. (g) Large mediastinal hematoma is easier to recog-nize after contrast material administration and causes mass effect on pulmonary arteries. (i–l) Contrast-enhanced transverse CT im-ages obtained after delay show that true and false lumina are now equally enhanced. Delayed enhancement of false lumen illustratesthat blood flow is slower in larger false lumen relative to smaller true lumen. (k) At 11 o’clock position, acute angle formed betweenanterior wall of aorta and intimomedial flap is beak sign (arrow), which is characteristic feature of false lumen.

(Figure on facing page.)

No

ntrau

matic A

ortic Em

ergen

cies

49

Figure 1. (caption on facing page).

Wes

t an

d B

hal

la

50

aorta, juxta–superior mesenteric artery aorta, juxtarenalaorta, and infrarenal aorta (11). Further, the reportshould identify any related complications, includingsigns of aortic rupture, aneurysm formation, or infarc-tion resulting from compromise of coronary, renal, ormesenteric arteries.

CT findings.—Helical CT is highly sensitive and spe-cific for the detection of acute aortic dissection (12).Direct CT signs include (a) detection of an intimo-medial flap (Figs 1, 2), (b) compression of the true lu-men by a contrast material–enhanced or unenhancedfalse lumen (Fig 1), and (c) displaced intimal calcifi-cation on unenhanced CT images (Fig 2d) (13). Lessspecific signs include widening of the aorta and thick-ening of the aortic wall.

When endovascular therapy is planned, distinguish-ing the true from the false lumen becomes important.In most cases, the true lumen is identified by its conti-nuity with the undissected portion of the aorta (14).Additional useful CT signs have been described (14).The beak sign, a feature of the false lumen, is an acuteangle formed between the intimomedial flap and theouter wall of the aorta. The beak may be white (filledwith contrast material) or gray (filled with hema-toma) (Fig 1k). Larger lumen size on cross-sectionalimages is a feature of the false lumen. In most cases,the true lumen collapses and is compressed by thelarger false lumen (Fig 1). Intimomedial rupture hasbeen depicted recently with multi–detector row CT

(15). In 8% of the cases, the free edges of the intimo-medial tear point from the true lumen into the falselumen (Fig 3). When multiple phases of scanning areperformed after the intravenous administration ofcontrast material, the false lumen may be noted tohave delayed enhancement and delayed washoutcompared with the true lumen (Fig 1).

Intramural Hematoma

Pathology.—The second member of the acute aorticsyndrome family, intramural hematoma, was first de-scribed in 1920 by Krukenburg (16) but was notwidely recognized in the English language radiologyliterature until the 1990s. Patients with intramural he-matoma present with the acute aortic syndrome, clini-cally indistinguishable from acute aortic dissection. In

Figure 2. Type B aortic dissection in a pa-tient with Marfan disease. (a) Contrast-en-hanced transverse CT image at origin of leftsubclavian artery shows proximal extent oflarge false lumen. (b) In distal descendingaorta, larger false lumen compresses ante-rior true lumen. (c) Contrast-enhanced and(d) unenhanced transverse CT images atsame level show some intimal calcificationfrom posterior wall displaced far anteriorlyalong intimomedial flap.

Figure 3. Type Aaortic dissectionwith intimomedialrupture. Contrast-enhanced trans-verse CT imageshows that freeedges of intimo-medial tear pointfrom true lumen onpatient’s left intofalse lumen onpatient’s right.

No

ntrau

matic A

ortic Em

ergen

cies

51

acute aortic dissection, the initial lesion is tearing ofthe intima. In contrast, the initial lesion in intramuralhematoma is rupture of the vasa vasorum, resulting inhemorrhage into the aortic media without tearing ofthe intima (5). Because the lesion is confined to theaortic wall, no false lumen develops. Systemic hyper-tension and blunt chest trauma are risk factors for de-velopment of intramural hematoma (17).

Intramural hematoma may progress to aortic dissec-tion. Intramural hematoma involving the ascendingaorta (Stanford type A) and the presence of pericardialeffusion are predictors of subsequent development ofaortic dissection (18). Because of this potential com-plication, intramural hematoma involving the ascend-ing aorta is usually treated surgically. Intramural he-matoma involving the descending aorta is usuallytreated medically. Intramural hematoma may regresswhen managed medically (18).

CT findings.—The major CT finding of intramuralhematoma is a hyperattenuating crescentic eccentri-cally located region of aortic wall thickening onunenhanced CT images (Fig 4) (13). The hematomamay compress the aortic lumen. The lesion may not

be detectable on contrast-enhanced CT images be-cause the unenhanced hematoma has nearly the sameattenuation as the aortic wall when displayed with thewindow setting sufficiently wide to display the high-attenuation contrast material in the aortic lumen. In-deed, the lack of enhancement helps to distinguish in-tramural hematoma from acute aortic dissection. In-tramural hematoma may resolve (Fig 4), may progressto acute aortic dissection, or may rupture (Fig 5).

Intramural hematoma should be distinguishedfrom mural thrombus, which is induced by thrombo-genic atheroma or alteration in laminar blood flow.Intramural hematoma is a subintimal process, whilemural thrombus occurs on the surface of the intima. Ifthe intima is calcified on unenhanced CT images, itsposition relative to the thrombus or hematoma allowseasy distinction between these two entities.

Penetrating Atherosclerotic Ulcer

Pathology.—The third member of the acute aortic syn-drome family, penetrating atherosclerotic ulcer, was de-scribed initially by Shennan in 1934 (19). Stanson andcolleagues (20) provided a more complete description

Figure 4. Type B intramural hematoma that resolved after 4 months. (a, b) Unenhanced transverse CT images show crescentichigh-attenuation hematoma on posteromedial aspect of descending thoracic aorta. (c, d) Contrast-enhanced transverse CT imagesshow that intramural hematoma is visible as thickened region of aortic wall. Smooth contour of contrast-filled lumen without intimo-medial tear and absence of enhancement help distinguish this intramural hematoma from type B aortic dissection. There are no inti-mal calcifications to help distinguish intramural hematoma from mural thrombus, but shape and long craniocaudal extent of pathologicfindings are typical of intramural hematoma. (e, f) Contrast-enhanced transverse CT images obtained after 4 months of medical therapyshow that wall of aorta is uniformly thin, without region of high attenuation, indicating that intramural hematoma has resolved.

Wes

t an

d B

hal

la

52

in 1986. Because of the relatively recent recognition ofpenetrating atherosclerotic ulcer, the literature definingits natural history and correlating imaging and patho-logic findings is limited.

Patients may present with the acute aortic syn-drome or may be asymptomatic. Penetrating athero-sclerotic ulcer is defined as an atheromatous ulcerthat violates the aortic intima and penetrates intothe media (5). In most cases, the ulcer precipitateshemorrhage (intramural hematoma) within the aor-tic wall. Localized aortic dissection may occur but islimited to the area surrounding the ulcer becausetransmural inflammation associated with penetrat-ing atherosclerotic ulcer fuses the layers of the aorticwall, which prevents dissection from extendingmuch beyond the ulcer. Penetrating atheroscleroticulcer may regress, may remain stable in size, mayenlarge to form an aneurysm or pseudoaneurysm, ormay rupture (21).

When compared with acute aortic dissection, pen-etrating atherosclerotic ulcer occurs in older patientswho have extensive atherosclerosis. Penetrating ath-erosclerotic ulcer occurs in the descending thoracicor abdominal aorta and is relatively uncommon inthe ascending aorta. Because penetrating atheroscle-rotic ulcer is a marker of a severely diseased aorta,surgery is usually reserved for a penetrating athero-sclerotic ulcer that has ruptured or one for whichmedical management has failed.

CT findings.—Calcified atheromatous plaques arereadily identified on unenhanced CT images. Ulcer-ation in the center of an atheromatous plaque fre-quently occurs in patients with advanced atherosclero-sis. The diagnosis of penetrating atherosclerotic ulceris reserved for those relatively rare instances in whicha contrast material–filled outpouching extends be-yond the plaque and into the wall of the aorta, andthe penetrating ulcer is usually surrounded by intra-mural hematoma (Fig 6) (22).

Aortic Aneurysm

Aortic aneurysm may occur anywhere along theaorta from the ascending aorta to the bifurcation. Atrue aortic aneurysm is defined as focal irreversible dila-tion of the aorta involving all three layers of the aorticwall. A false aortic aneurysm (pseudoaneurysm) is a fo-cal irreversible dilation of the aorta that does not in-volve all three layers of the wall. Atherosclerotic aneu-rysm is the most common example of a true aneu-rysm. In contrast, traumatic pseudoaneurysm involvestearing of the intima with a focal bulging of the aorticwall contained by the adventitia and possibly a por-tion of the media.

Aneurysms are described according to their shape(fusiform or saccular), location (ascending, descend-ing, or abdominal), and relationship to importantvessels (left subclavian, renal, and iliac arteries). Aneu-rysms are diagnosed when the luminal diameter of

Figure 5. Ruptured intramural hematoma.(a–c) Unenhanced transverse CT imagesthrough descending thoracic aorta demon-strate high-attenuation crescentic region ofthickened aortic wall anteriorly and laterally.(c) High-attenuation area extends into an-eurysm in distal descending thoracic aorta.Adjacent mediastinal hematoma and largeright hemothorax are signs of aortic rupture.(d) Conventional aortogram in left anterioroblique projection shows aneurysm only.Because only the lumen is evaluated withconventional aortography, intramural he-matoma is not visible.

No

ntrau

matic A

ortic Em

ergen

cies

53

the abnormal segment exceeds the diameter of an ad-jacent normal segment by at least 50%. Focal dilationof the thoracic aorta to 5 cm or more and dilation ofthe abdominal aorta to 3 cm or more are the generallyaccepted criteria for diagnosing an aneurysm. Most ath-erosclerotic aneurysms are fusiform and often have asmooth contour. In contrast, most infected (mycotic)aneurysms are saccular, with a lobulated contour (23).The pathogenesis of atherosclerotic aortic aneurysms isspeculative, but loss of elastin from the media and in-ner one-third of the adventitia probably plays an im-portant role (24).

The etiology of aortic aneurysm is multifactorial, withgenetic, environmental, and physiologic factors eachcontributing (24). In the ascending aorta, congenitaldisorders such as the Marfan syndrome and the Ehlers-Danlos syndrome are the main causes. Chronic aorticdissection is another causative factor in the develop-ment of ascending aortic aneurysm. In the descendingthoracic aorta and the abdominal aorta, hypertensionand atherosclerosis are the main risk factors. Infection(mycotic aneurysm and syphilitic aneurysm) and in-flammation are relatively uncommon causes.

The typical thoracic aortic aneurysm occurs in thedescending thoracic aorta, is fusiform in shape, andoccurs in an elderly man with hypertension and ath-erosclerosis. The typical abdominal aortic aneurysm isinfrarenal in location and fusiform in shape and oc-curs in the same type of patient.

Many aortic aneurysms are asymptomatic and arediscovered incidentally with imaging studies, with pa-tient participation in a health screening program, or atphysical examination. These asymptomatic aneurysmsare often followed periodically with US, CT, or MRimaging to assess their size and rapidity of enlarge-ment with time. Imaging of asymptomatic aneurysmsis intended to identify risk factors for impending rup-ture so that surgical or endovascular repair may be un-dertaken just in time (ie, prior to rupture). That is, re-pair should be timed to occur before the aneurysmruptures but sufficiently close to the anticipated time

of rupture that the risk of repair is less than the risk ofrupture and resultant death (25).

The diameter of the aneurysm is the easiest param-eter to follow over time. In one large series, the me-dian diameter at which rupture or dissection compli-cated an atherosclerotic thoracic aortic aneurysm was6.0 cm in the ascending aorta and 7.2 cm in the de-scending thoracic aorta (26). Therefore, the authors ofthat series recommend preemptive surgical therapy foratherosclerotic aneurysms at a diameter of 5.5 cm forascending aortic aneurysms and 6.5 cm for descend-ing thoracic aneurysms (26). For patients with Marfansyndrome, the parameters are a diameter of 5.0 cm forascending aortic aneurysms and 6.0 cm for descend-ing thoracic aneurysms (26). In the abdominal aorta,5.0 cm is a widely quoted threshold for elective repair.Other indications for surgical repair include rupture,acute aortic dissection, pain consistent with rupture,compression on adjacent organs (trachea, esophagus,bronchi), or development of aortic insufficiency (26).Growth of the aneurysm at a rate of 1 cm/y or more isalso an indication for repair (26,27).

More sophisticated morphologic parameters, suchas CT estimation of wall stress, may provide betterpredictors of the risk of aneurysm rupture (28). Thisadvanced technique requires substantial computingpower and the thinner images routinely obtained withmulti–detector row CT scanners.

Patients with aortic aneurysm come to the emer-gency center when they have pain related to aortic an-eurysm. Symptoms develop when the aortic aneurysmcreates a mass effect on an adjacent structure, enlargesrapidly, or ruptures. In an emergency situation, CTscanning without intravenous contrast material maybe used to establish the presence and measure the sizeof an aortic aneurysm.

Unenhanced CT will also readily depict signs of aor-tic rupture or impending aortic rupture. The principaldiagnostic sign of aortic rupture is periaortic hema-toma (Fig 7). In the thorax, mediastinal hematoma orhemothorax and, in the abdomen, retroperitoneal

Figure 6. Penetrating atherosclerotic ulcerin aortic arch. (a) Unenhanced and (b) con-trast-enhanced transverse CT images ataortic arch show intramural hematoma withintimal calcifications on surface of lumen. Ifthis were mural thrombus, intimal calcifica-tions would have been external to thrombus.(b) Contrast-filled ulcer (arrow) penetratesinto media and is surrounded by intramuralhematoma.

Wes

t an

d B

hal

la

54

hematoma are easily recognized without intravenouscontrast material administration. In the absence ofperiaortic hematoma, demonstration of a well-definedcrescent-shaped area of increased attenuation withinthe aortic wall indicates impending rupture (29–33).This hyperattenuating crescent sign represents anacute intramural or mural thrombus hemorrhage andis a CT sign of acute or impending rupture. Demon-stration of a focal defect in an otherwise calcified aor-tic wall is another sign of impending aortic rupture(Fig 8) (32).

When the clinical situation permits, intravenous ad-ministration of contrast material permits detailedmorphologic analysis of the aortic aneurysm, particu-larly when a multi–detector row CT scanner is used.CT angiography demonstrates the anatomic structuresof complex aortic aneurysms and shows the relation-ship of the aneurysm to adjacent vessels (34). Thepresence and extent of mural thrombus are readily ap-parent when the aortic lumen is enhanced. In the ab-domen, the number and location of renal arteries canbe established to aid in planning treatment. CT an-giography helps to determine whether or not an aneu-

rysm is suitable for treatment with endovascular stent-graft placement (35–37).

Aortic Occlusion

Acute occlusion of the abdominal aorta causes pain,pallor, pulselessness, paresthesia, and paralysis of thelower extremities. Either embolism or in situ throm-bosis may cause the occlusion. Paresthesia and paraly-sis indicate limb ischemia, requiring urgent embolec-tomy or revascularization (38). CT can be used toidentify the level of aortic occlusion but is limited inits ability to show the extent of blockage, which is bet-ter assessed with conventional aortography (38).

CT TECHNIQUE

The CT scanning technique and contrast injectiontechnique may be tailored to imaging of the aorta.The use of a general chest, abdomen, and pelvis proto-col, which is helpful for evaluation of trauma, limitsthe spatial resolution and arterial enhancement thatcan be achieved by optimizing the parameters specifi-cally for aortic imaging.

Figure 7. Ruptured abdominal aortic aneurysm. (a, b) Unenhanced transverse CT images of abdominal aorta show contained he-matoma in right retroperitoneum at 7 o’clock position. (c) Contrast-enhanced transverse CT image at same level as in b shows ex-tensive mural thrombus within abdominal aortic aneurysm. Note intimal calcifications external to mural thrombus at 11, 1, and 5o’clock positions.

Figure 8. Leaking abdominal aortic aneu-rysm. (a, b) Contrast-enhanced transverseCT images. (a) Image obtained in midpor-tion of abdomen demonstrates completering of intimal calcification. Large muralthrombus separates intima from contrast-enhanced lumen. (b) More caudally, intimalcalcifications are disrupted at 11 o’clock po-sition, with hematoma extending from pointof aortic rupture into right retroperitoneum.

No

ntrau

matic A

ortic Em

ergen

cies

55

Obtaining an unenhanced scan prior to the CTangiography is recommended by several authors(13,39,40). Others use only contrast-enhanced ex-aminations (34,41). Advocates of unenhanced imag-ing emphasize its value for the detection of intramu-ral hematoma and displaced or disrupted intimalcalcifications, CT findings that are often obscured byhigh-attenuation contrast material in the aortic lu-men. If unenhanced CT is to be performed, obtain-ing 5-mm images with a maximum gantry rotationspeed, a radiation-efficient detector configuration(4 × 3.75 mm, 4 × 5 mm, 8 × 2.5 mm, or 16 × 1.25mm), and relatively high table speed (pitch > 1.0:1),should limit radiation exposure while producing ac-ceptable images. As discussed in the section on “Aor-tic Aneurysm,” a rapid unenhanced CT examinationmay be the first and only imaging procedure re-quired if aortic rupture is suspected and if CT signsof rupture are unequivocally present.

CT aortography may be performed by using stan-dard protocols for contrast material injection, such as150 mL of nonionic contrast material (300 mg of io-dine per milliliter) injected at a rate of at 3.0 mL/secafter a 30-second delay. This standard injection proto-col offers the advantage of simplicity. It is a one-size-fits-all approach that will yield adequate but clearlynot optimal results. Such an approach is particularlyvaluable in busy emergency centers, where tailoringcontrast material administration to specific clinical in-dications and patient factors may reduce efficiency orincrease the likelihood of an error. Although we in-tuitively assume that production of the best-qualityimages results in the best diagnoses, we know of noscientific study that evaluates this hypothesis. Insome settings, the one-size-fits-all approach may bepreferable.

For optimal aortic imaging, the technical details ofcontrast material administration are critical. In gen-eral, high iodine concentrations (300–400 mg of io-dine per milliliter) and rapid injection rates (4–6 mL/sec) are needed. The delay between onset of injectionand onset of scanning should be customized to eachpatient by using either the test-bolus or automatic scan-triggering techniques available on all multi–detectorrow CT scanners. The duration of contrast material in-

jection should be adjusted so that injection continuesthrough the entire scan of the aorta but stops whenscanning is completed. The use of a saline flush at theend of contrast material injection offers theoretical ad-vantages in reducing the contrast dose but is not easyto implement in routine clinical practice. For a de-tailed discussion of techniques of contrast material in-jection, the reader should consult an excellent articleby Fleischmann (42).

CT scanner settings for CT angiography are necessar-ily specific to the vendor and model of the scanner. Onsingle–detector row scanners, the fastest possible gantryrotation time (0.75–1.0 second), moderately thin colli-mation (3–5 mm), and high pitch (2.0:1) producegood results (43). On four–detector row scanners, thechoice of detector configuration will depend on thelength of the torso to be scanned. To image the entirethoracoabdominal aorta, including the origins of aorticbranches and the iliac arteries, with a four–detector rowCT scanner, the 4 × 2.5-mm mode with high pitch(1.5:1) is needed. For shorter scans, the 4 × 1.25-mmmode may be used. On eight– and 16–detector rowscanners, the 8 × 1.25-mm or 16 × 1.25-mm mode, witha pitch of 1.35:1 on both eight– and 16–detector rowscanners or a pitch of 0.67:1 on 16–detector row scan-ners, should produce excellent results. Obtaining thin-ner images in the 16 × 0.625-mm mode is possible, butno published literature is available regarding the utilityof submillimeter imaging for aortic abnormalities.

Diagnosis may be made from 5-mm transverse CTimages. Thinner 2.5-mm images may be useful in se-lected areas, if finer detail is needed for equivocalfindings. Routine review of thin, noisy transverse CTimages (0.625–2.5 mm) of the entire aorta is both te-dious and impractical.

Transverse CT images are sufficient for diagnosis ofaortic abnormalities and are probably the only imagesthe emergency radiologist needs to establish the diag-nosis. Treatment planning may be aided by analysis ofpostprocessed images. Two- and three-dimensionalmaximum intensity projections, curved planar refor-mations, and multiplanar volume-rendered imagesdepict the gross morphologic structure of the diseasedaorta and the relationship to nearby vessels. If thesepostprocessing techniques are used, obtaining over-lapping transverse images improves image quality. A2:1 overlap is frequently recommended, but the appli-cability of this degree of overlap to submillimeter im-ages has not been established.

PITFALLS

One of the most problematic artifacts encountered inaortic imaging is pulsation of the ascending aorta,which may simulate the appearance of acute aorticdissection or intramural hematoma (Fig 9). Alterna-tively, aortic pulsations may mask real pathologic

Figure 9. Motionartifact. Contrast-enhanced trans-verse CT imageshows linear bandof low attenuationalong anterolateralwall of aortic archthat might be mis-taken for intimo-medial flap.

Wes

t an

d B

hal

la

56

findings. To some extent, the experienced radiologistcan learn to read around these artifacts, but a more de-finitive solution is needed. The current generation of16–detector row (or higher) CT scanners with gantryrotation times of 500 msec or less may allow routineuse of electrocardiographic gating to substantially re-duce aortic pulsation artifacts (44).

Technical problems often degrade image quality.Bolus timing errors may be reduced by using test in-jection or automated scan-triggering tools. Streak arti-facts may be reduced, but not eliminated, by movingthe upper extremities above the head and by minimiz-ing metal hardware on the torso of the patient. Radi-ologists become adept at recognizing streak artifacts aslinear low-attenuation bands on one or a few adjacentimages, usually emanating from an adjacent high-at-tenuation structure. Streaks from the contrast-en-hanced superior vena cava or left brachiocephalic veinare familiar sources of image degradation. To a lim-ited extent, streak artifacts may be reduced by usingstrategies to dilute contrast material, to add a salinechaser, to inject the right rather than the left arm, or toinject the lower rather than the upper extremity.

Normal periaortic anatomic structures may be mis-taken for aortic dissection. The origins of the aortic archbranches and the left brachiocephalic, superior inter-costal, or pulmonary veins may mimic the appearanceof a double lumen (45). The thymus, superior pericar-dial recess, atelectatic lung, enlarged lymph nodes,pleural thickening, or pleural effusion may lead the un-wary into a false diagnosis of aortic abnormalities.

In conclusion, CT, particularly multi–detector rowhelical CT, is an excellent tool for evaluating the aortafor dissection, intramural hematoma, penetrating ath-erosclerotic ulcer, or aortic aneurysm. Patients with aor-tic abnormalities often present to the emergency centerwith the acute aortic syndrome, which is characterizedby severe, intense, acute searing or tearing, throbbing,and migratory chest pain. An understanding of the pa-thology of aortic emergencies, particularly the under-standing that failure of the aortic media is common toall of these entities, helps the radiologist to make senseof imaging findings. Unenhanced CT is often valuableand may be the only imaging study required for diag-nosis. Contrast-enhanced CT studies performed duringpeak aortic enhancement allow excellent transverse im-ages of the aorta and are usually sufficient to diagnoseand fully characterize the aortic abnormalities. Manypostprocessing techniques are available and may aid intreatment planning, but they have not replaced trans-verse CT images as the primary diagnostic tool for theemergency radiologist.

References1. Sommer T, Fehske W, Holzknecht N, et al. Aortic dissec-

tion: a comparative study of diagnosis with spiral CT,multiplanar transesophageal echocardiography, and MR im-aging. Radiology 1996; 199:347–352.

2. Vilacosta I, Roman JA. Acute aortic syndrome. Heart 2001;85:365–368.

3. Wooley CF, Sparks EH, Boudoulas H. Aortic pain. ProgCardiovasc Dis 1998; 40:563–589.

4. Klompas M. Does this patient have an acute thoracic aorticdissection? JAMA 2002; 287:2262–2272.

5. Coady MA, Rizzo JA, Elefteriades JA. Pathologic variantsof thoracic aortic dissections: penetrating atherosclerotic ul-cers and intramural hematomas. Cardiol Clin 1999; 17:637–657.

6. Hirst AE Jr, Johns VJ Jr, Kime SW Jr. Dissecting aneurysmof the aorta: a review of 505 cases. Medicine (Baltimore)1958; 37:217–279.

7. Larson EW, Edwards WD. Risk factors for aortic dissection:a necropsy study of 161 cases. Am J Cardiol 1984; 53:849–855.

8. Farber A, Wagner WH, Cossman DV, et al. Isolated dissec-tion of the abdominal aorta: clinical presentation and thera-peutic options. J Vasc Surg 2002; 36:205–210.

9. Daily PO, Trueblood HW, Stinson EB, Wuerflein RD,Shumway NE. Management of acute aortic dissections. AnnThorac Surg 1970; 10:237–247.

10. DeBakey ME, Henly WS, Cooley DA, Morris GC Jr,Crawford ES, Beall AC Jr. Surgical management of dissect-ing aneurysms of the aorta. J Thorac Cardiovasc Surg 1965;49:130–149.

11. Quint LE, Platt JF, Sonnad SS, Deeb GM, Williams DM.Aortic intimal tears: detection with spiral computed tomogra-phy. J Endovasc Ther 2003; 10:505–510.

12. Yoshida S, Akiba H, Tamakawa M, et al. Thoracic involve-ment of type A aortic dissection and intramural hematoma:diagnostic accuracy—comparison of emergency helical CTand surgical findings. Radiology 2003; 228:430–435.

13. Ledbetter S, Stuk JL, Kaufman JA. Helical (spiral) CT in theevaluation of emergent thoracic aortic syndromes: traumaticaortic rupture, aortic aneurysm, aortic dissection, intramuralhematoma, and penetrating atherosclerotic ulcer. RadiolClin North Am 1999; 37:575–589.

14. LePage MA, Quint LE, Sonnad SS, Deeb GM, Williams DM.Aortic dissection: CT features that distinguish true lumenfrom false lumen. AJR Am J Roentgenol 2001; 177:207–211.

15. Kapoor V, Ferris JV, Fuhrman CR. Intimomedial rupture: anew CT finding to distinguish true from false lumen in aorticdissection. AJR Am J Roentgenol 2004; 183:109–112.

16. Krukenburg E. Baitrage zur frage des aneurysma dis-secans. Beitr Pathol Anat 1920; 67:329–351.

17. Alfonso F, Goicolea J, Aragoncillo P, Hernandez R, MacayaC. Diagnosis of aortic intramural hematoma by intravascularultrasound imaging. Am J Cardiol 1995; 76:735–738.

18. Choi SH, Choi SJ, Kim JH, et al. Useful CT findings for pre-dicting the progression of aortic intramural hematoma toovert aortic dissection. J Comput Assist Tomogr 2001; 25:295–299.

19. Shennan T. Dissecting aneurysms. Medical ResearchCouncil Special Report series. London, United Kingdom:Medical Research Council, 1934.

20. Stanson AW, Kazmier FJ, Hollier LH, et al. Penetrating ath-erosclerotic ulcers of the thoracic aorta: natural history andclinicopathologic correlations. Ann Vasc Surg 1986; 1:15–23.

21. Cho KR, Stanson AW, Potter DD, Cherry KJ, Schaff HV,Sundt TM. Penetrating atherosclerotic ulcer of the descend-ing thoracic aorta and arch. J Thorac Cardiovasc Surg2004; 127:1393–1401.

22. Macura KJ, Corl FM, Fishman EK, Bluemke DA. Pathogen-esis in acute aortic syndromes: aortic dissection, intramuralhematoma, and penetrating atherosclerotic aortic ulcer.AJR Am J Roentgenol 2003; 181:309–316.

No

ntrau

matic A

ortic Em

ergen

cies

57

23. Macedo TA, Stanson AW, Oderich GS, Johnson CM,Panneton JM, Tie ML. Infected aortic aneurysms: imagingfindings. Radiology 2004; 231:250–257.

24. Coady MA, Rizzo JA, Goldstein LJ, Elefteriades JA. Naturalhistory, pathogenesis, and etiology of thoracic aortic aneu-rysms and dissections. Cardiol Clin 1999; 17:615–635; vii.

25. Juvonen T, Ergin MA, Galla JD, et al. Prospective study ofthe natural history of thoracic aortic aneurysms. Ann ThoracSurg 1997; 63:1533–1545.

26. Coady MA, Rizzo JA, Elefteriades JA. Developing surgicalintervention criteria for thoracic aortic aneurysms. CardiolClin 1999; 17:827–839.

27. Scott RA, Tisi PV, Ashton HA, Allen DR. Abdominal aorticaneurysm rupture rates: a 7-year follow-up of the entire ab-dominal aortic aneurysm population detected by screening.J Vasc Surg 1998; 28:124–128.

28. Fillinger MF, Marra SP, Raghavan ML, Kennedy FE. Predic-tion of rupture risk in abdominal aortic aneurysm during ob-servation: wall stress versus diameter. J Vasc Surg 2003;37:724–732.

29. Pillari G, Chang JB, Zito J, et al. Computed tomography ofabdominal aortic aneurysm: an in vivo pathological reportwith a note on dynamic predictors. Arch Surg 1988; 123:727–732.

30. Posniak HV, Olson MC, Demos TC, Benjoya RA, MarsanRE. CT of thoracic aortic aneurysms. RadioGraphics 1990;10:839–855.

31. Mehard WB, Heiken JP, Sicard GA. High-attenuating cres-cent in abdominal aortic aneurysm wall at CT: a sign ofacute or impending rupture. Radiology 1994; 192:359–362.

32. Siegel CL, Cohan RH, Korobkin M, Alpern MB, CourneyaDL, Leder RA. Abdominal aortic aneurysm morphology: CTfeatures in patients with ruptured and nonruptured aneu-rysms. AJR Am J Roentgenol 1994; 163:1123–1129.

33. Gonsalves CF. The hyperattenuating crescent sign. Radiol-ogy 1999; 211:37–38.

34. Hartnell GG. Imaging of aortic aneurysms and dissection:CT and MRI. J Thorac Imaging 2001; 16:35–46.

35. Thurnher SA, Grabenwoger M. Endovascular treatment ofthoracic aortic aneurysms: a review. Eur Radiol 2002; 12:1370–1387.

36. Fattori R, Napoli G, Lovato L, et al. Descending thoracicaortic diseases: stent-graft repair. Radiology 2003; 229:176–183.

37. Schoder M, Cartes-Zumelzu F, Grabenwoger M, et al. Elec-tive endovascular stent-graft repair of atherosclerotic tho-racic aortic aneurysms: clinical results and midterm follow-up. AJR Am J Roentgenol 2003; 180:709–715.

38. Surowiec SM, Isiklar H, Sreeram S, Weiss VJ, LumsdenAB. Acute occlusion of the abdominal aorta. Am J Surg1998; 176:193–197.

39. Bhalla S, Menias CO, Heiken JP. CT of acute abdominalaortic disorders. Radiol Clin North Am 2003; 41:1153–1169.

40. Castaner E, Andreu M, Gallardo X, Mata JM, CabezueloMA, Pallardo Y. CT in nontraumatic acute thoracic aorticdisease: typical and atypical features and complications.RadioGraphics 2003; 23(special issue):S93–S110.

41. Rubin GD. MDCT imaging of the aorta and peripheral ves-sels. Eur J Radiol 2003; 45(suppl 1):S42–S49.

42. Fleischmann D. Use of high concentration contrast media:principles and rationale—vascular district. Eur J Radiol2003; 45(suppl 1):S88–S93.

43. Rubin GD, Shiau MC, Leung AN, Kee ST, Logan LJ, SofilosMC. Aorta and iliac arteries: single versus multiple detector-row helical CT angiography. Radiology 2000; 215:670–676.

44. Roos JE, Willmann JK, Weishaupt D, Lachat M, MarincekB, Hilfiker PR. Thoracic aorta: motion artifact reduction withretrospective and prospective electrocardiography-assistedmulti–detector row CT. Radiology 2002; 222:271–277.

45. Batra P, Bigoni B, Manning J, et al. Pitfalls in the diagnosisof thoracic aortic dissection at CT angiography.RadioGraphics 2000; 20:309–320.

NO

TES

59

Cardiac Applications forMulti–Detector Row CT in the

Emergency Department1

Udo Hoffmann, MD, Ricardo C. Cury, MD, Maros Ferencik, MD, PhD,Fabian Moselewski, BS, Suhny Abbara, MD, and Thomas J. Brady, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 59–69.

1From the Department of Radiology, Massachusetts General Hospital, 100 Charles River Plaza, Suite 400, Boston, MA02114 (e-mail: uhoffman@partners.org).

Although more than 6 million patients present with acute chest pain to emergency de-partments in the United States each year and roughly a third of them are admitted tothe hospital, only a fraction (∼10%) are subsequently diagnosed with acute coronarysyndrome. This practice can be attributed to the low sensitivity of early biomarkers(troponin) and changes in the electrocardiogram (ECG) for acute coronary syndrome.In addition, missed acute myocardial infarctions are still responsible for 20% of emer-gency department malpractice dollar losses (1,2), and the number of missed myocar-dial infarctions remains relatively high (1%–3%), albeit decreased from approxi-mately 6% many years ago (1,3,4).

Despite the fact that significant coronary artery disease (>50% coronary stenosis) isthe leading cause of acute coronary syndrome (90%) in patients with acute chest pain,current strategies to diagnose acute coronary syndrome in the emergency departmentdo not include morphologic information on the presence and severity of coronary ar-tery disease. Although cardiac multi–detector row computed tomography (CT) is not apart of the usual clinical care in patients with chest pain, it is conceivable that the fastand noninvasive detection of the presence or absence of significant coronary artery ste-nosis constitutes an attractive approach to substantially improve the clinical care ofpatients with acute chest pain.

DIAGNOSTIC STRATEGY FOR PATIENTS WITH ACUTE CHEST PAIN

Patients who present with typical clinical symptoms without initial elevation of enzymelevels who have a nondiagnostic ECG are observed for 6–12 hours while repeat ECGand biomarker tests are performed. Patients may stay in the hospital for 24 hours orlonger until a negative stress test has been obtained to exclude acute coronary syndrome.

Stress testing is routinely used in clinical practice to assess patients for inducible is-chemia, but such testing has limitations. In a population with a 20% prevalence of sig-nificant coronary artery disease, exercise tolerance tests and scintigraphic myocardialstudies (technetium Tc 99m sestamibi nuclear perfusion imaging or stress [exercise ordobutamine] echocardiography) have good sensitivity for the detection of significantunderlying coronary artery disease (sensitivity of 76% for the exercise tolerance test,83% for technetium Tc 99m sestamibi nuclear perfusion imaging, and 85% for stressechocardiography). With the exception of stress echocardiography, these same three

Ho

ffm

ann

et

al

60

tests have relatively poor specificity (60%, 64%, and77%, respectively) for the detection of significant un-derlying coronary artery disease (5–7). In addition,stress tests can only be performed after acute myocar-dial infarction has been excluded (usually excluded bynegative results on two sets of serum troponin assays),and stress tests are complex procedures, requiring time(ie, approximately 150 minutes for stress single pho-ton emission CT) and expertise, which is not usuallyavailable 24 hours per day and 7 days per week.

Thus, accurate noninvasive detection of significant cor-onary artery stenosis in the emergency department set-ting would have an incremental value to current proce-dures: (a) Patients with chest pain and hemodynami-cally significant underlying coronary artery disease couldbe identified more quickly and treated earlier and thushave potentially improved clinical outcomes. (b) Thenumber of unnecessary hospital admissions could be re-duced, and the efficacy and cost-effectiveness of the careof patients with chest pain could be improved.

MULTI–DETECTOR ROW CT TECHNIQUE ANDPROTOCOLS

With short examination times and robust image qual-ity, cardiac CT imaging constitutes a highly attractiveapproach for patient work-up in the setting of theemergency department. In addition, most CT scannersnow provide easy handling of the large number of CTimages, including image reconstruction at differentphases of the cardiac cycle and postprocessing. Thus,diagnostic evaluation of the findings from an exami-nation can usually be completed within minutes ofcompletion of the scanning procedure. However,proper selection and preparation of patients are re-quired to optimize diagnostic image quality.

Patient Selection and Preparation

In candidates for CT coronary angiography, relativecontraindications are (a) a previous severe reaction toan iodinated contrast agent or (b) renal failure (serumcreatinine level, >1.5 mg/dL [>133 µmol/L]). To en-sure diagnostic image quality, the heart of the patientshould be in sinus rhythm, and a target heart rate ofless than 65 beats per minute should be achieved withintravenous administration of short-acting β-adrenergicblocking agents (ie, 5 mg of metoprolol). In most pa-tients in whom the target heart rate cannot be achieved,image quality is suboptimal, and the diagnostic valueof the CT scan is limited. In addition, patients shouldbe able to perform a breath hold of 15–25 seconds.

A heart rate of less than 65 beats per minute is almostimperative for diagnostic image quality. In our institu-tion, patients receive a short-acting β-adrenergic blockingagent intravenously immediately prior to the examina-tion unless their heart rate is less than 60 beats perminute or contraindications are present (eg, congestive

heart failure, asthma, bradycardia, atrioventricularblock). The necessity for β-adrenergic blocking agent ad-ministration can be assessed during a test breath hold.Usually, the heart rate drops 5–10 beats per minute dur-ing the first 30 seconds of an inspirational breath hold.

The patient is placed in the supine position in thescanner, and three ECG leads are attached. An appro-priately sized intravenous line is placed into the an-tecubital vein, although most patients from the emer-gency department already have intravenous access es-tablished. The level of monitoring of those patientstypically includes blood pressure, heart rate, ECG, andoxygen saturation, which should be maintainedthroughout the transport and imaging procedure.

The multi–detector row CT imaging protocol forcoronary angiography consists of three steps:

1. Localization.—The heart position is localized in atopographic scan of the chest.

2. Determination of contrast agent transit time.—Ten mil-liliters of a contrast agent (eg, iodixanol [320 mg of io-dine per milliliter]), immediately followed by 40 mL ofsaline (optional), is injected at a flow rate of 4 mL/sec.Ten seconds after initiation of the contrast agent injec-tion, axial images are acquired at the level of the aorticroot (3.0-mm collimation), followed by subsequent im-ages acquired at the same level at intervals of 2 seconds.Images are instantly displayed, and imaging is termi-nated when sufficient contrast enhancement of the aorticlumen is detected. The time interval from initiation ofinjection to the peak opacification of the ascending aortarepresents the transit time of the contrast agent.

3. Data acquisition.—Images are acquired in a spiralmode during injection of 80–100 mL of contrastagent, according to the scan duration, followed by 40mL of saline solution (optional), at a rate of 4 mL/sec.The start of the image acquisition is delayed accordingto the previously determined contrast agent transittime. Images are acquired with a pitch of 2.8–3.4 mmper rotation. Typically, images are reconstructed witha 1-mm section thickness and a 0.5-mm overlap (16-section multi–detector row CT). Retrospectively, ECG-gated half-scan reconstruction is performed.

Raw CT data may be archived on a magneto-opticaldrive or a DVD. However, retrospective reconstruction ofdata sets can be performed online immediately after thescanning procedure on the console of the CT scanner.

Image Reconstruction and Postprocessing

To minimize motion artifacts, the image reconstruc-tion algorithm retrospectively synchronizes the imageinformation with the ECG tracing that is recorded si-multaneously. Usually, images are reconstructed at dif-ferent phases of the cardiac cycle. In many cases, imagesreconstructed in diastole (between 55% and 65% of theR-R interval) have the fewest motion artifacts, while insome cases, the right coronary artery is better depictedbetween 35% and 50% of the R-R interval.

Card

iac Ap

plicatio

ns fo

r Mu

lti–Detecto

r Row

CT

61

To achieve adequate temporal resolution, only in-formation for 180° plus the fan angle is used for im-age reconstruction, which results in a temporal resolu-tion of 210 msec in the center of rotation (gantry rota-tion time [Trot] is 0.42 second). A linear interpolationis performed between the data of those two detectorsthat are closest to the chosen image plane. At heartrates less than 65 beats per minute, only one heartcycle is used to generate an image, and the temporalresolution is Trot/2, assuming that the patient is posi-tioned in the center of the gantry. At higher heartrates, information from different detectors obtainedduring subsequent heart cycles (M) (as many as three)are used for image reconstruction, which improves thetemporal resolution to Trot/(M • 2). However, this al-gorithm requires an extremely stable heart rate. Inmost cases, beat-to-beat variation causes spatial blur-ring and impaired image quality.

The 16 central rows of the scanner detector define 12detectors or 16 detectors with 0.75-mm collimated sec-tion width (12 × 0.75-mm or 16 × 0.75-mm, respec-tively), depending on the software version. To limit im-age noise, images may be reconstructed by using a sec-tion thickness of 1 mm and an overlap of 0.5 mm.

The resulting data set primarily consists of overlap-ping axial two-dimensional images. Because of a nearlyisotropic resolution and voxel size (0.6 mm × 0.5 mm× 0.5 mm), any arbitrary plane can be calculated with-out a substantial loss of image information. Dependingon the application software, many options are availableto depict and display the acquired data. To depict spe-cific regions of interest (such as the origin of the leftcoronary artery), anatomically adjacent structures in-cluding the pulmonary trunk, the bones of the thorax,and the left atrial appendix, have to be digitally re-moved. Several functions permit moving and croppingor punching of the volume data set to most clearlyshow the important region and the anatomic structuresof interest (Fig 1). Here are some typical postprocessingtechniques to illustrate findings.

Multiplanar reformations.—The image plane can bechosen arbitrarily. Multiplanar reformations are most of-ten used to generate cross-sectional images of coronaryarteries or typical views, such as short- or long-axis views.

Curved multiplanar reconstructions.—This algorithmcreates a single image that displays the course of a coro-nary artery (section follows the centerline of the vessel).This display is most useful to demonstrate the course of

Figure 1. Images obtainedwith different techniques in apatient with significant stenosisof proximal LAD depicted withmulti–detector row CT and se-lective coronary angiography.(a) Axial thin-section (5-mm)MIP multi–detector row CT im-age shows significant narrowingof vessel lumen, with residualcontrast agent filling in presenceof eccentric noncalcified plaque(arrow). (b) Curved multiplanarreconstruction along centerlineof LAD demonstrates contrastenhancement of mid and distalsegments of artery. Arrow indi-cates noncalcified atherosclerot-ic plaque. (c) Volume-renderedthree-dimensional image (sur-face shadowing display) of heartdemonstrates tomographic viewof lesion (arrow). (d) Invasiveselective coronary angiographicimage demonstrates eccentricstenosis of proximal LAD, withresidual filiform lumen (arrow).Quantitative coronary angiogra-phy determined 94% luminal ob-struction.

Ho

ffm

ann

et

al

62

a coronary artery from the ostium to its distal end.Curved multiplanar reconstructions have been shownto be useful in the evaluation of contrast agent–en-hanced electron-beam CT scans of coronary arteries (8).

Maximal intensity projection.—Reconstructions canbe displayed as a single image or as a stack of imagessummarizing the information from adjacent sections.This results in a loss of spatial detail but improvementin image contrast and a decrease in image noise. Im-ages are similar to conventional angiograms. Maxi-mum intensity projections (MIPs) allow depiction ofa longer length of coronary artery lumen and havebeen shown to be more accurate for the depiction ofsignificant coronary artery lesions than multiplanarreconstructions and three-dimensional displays (9).MIPs are also useful for demonstration purposes.

Volume rendering.—Volume rendering includes the en-tire volume of data, sums the contributions of each voxelalong a line from the eye of the viewer through the dataset, and displays the resultant composite for each pixel ofthe display. By using crop and punch functions of mod-ern workstations, three-dimensional images of the heartcan be generated within minutes. The volume-rendereddisplay has been shown to be useful in showing the ana-tomic structures and the spatial relationships of the heartand its surrounding vascular system and may be usefulfor surgical planning (9,10).

Radiation Exposure in CT Coronary Angiography

The effective radiation exposure varies from 7.6 to6.7 mSv in male subjects and from 9.2 to 8.1 mSv infemale subjects and is comparable to that from rou-tine CT of the thorax. By using an ECG-controlleddose modulation that reduces the tube current duringsystole, exposure can be reduced effectively to ap-proximately 4.3 mSv. However, studies to demon-strate the feasibility of this technique are warranted. Incomparison, diagnostic selective conventional coro-nary angiography has a mean effective radiation doseof approximately 5 mSv.

Depiction of Coronary Artery Calcification: CTScanning Protocol

The established imaging protocol for coronary calcifi-cation uses sequential image acquisition with prospec-tive ECG triggering. Considerations with respect to heartrate and image quality are similar to those for retrospec-tive imaging. However, the literature lacks studies dem-onstrating the effect of β-adrenergic blocking agents.

Scanning is performed by using prospective ECGtriggering during a single breath hold (12 seconds)and sequential data acquisition. Scans are prospec-tively initiated at 50% of the R-R interval. The scantypically results in 48 consecutive nonoverlapped 2.5-mm-thick sections (120–140 kVp, 150 mAs, temporalresolution of nearly 330 msec for a gantry rotationtime of 500 msec [240° partial scan]). The effective

patient dose is 1.0 mSv. Nonenhanced spiral retro-spectively ECG-gated protocols for the depiction ofcoronary calcium are also available. With less than 3-mm collimation and the possibility of multiple recon-structions within the R-R interval, these protocols mayyield better sensitivity and reproducibility but at thecost of higher radiation dose.

IMAGE ANALYSIS: DETECTION OF SIGNIFICANTCORONARY STENOSIS

Usually, axial and multiplanar reformatted cross-sec-tional images reconstructed perpendicular to the vesselcenterline and MIPs are used to assess the presence ofsignificant luminal obstruction. The presence of signifi-cant luminal obstruction is visually assessed. In our ex-perience, significant stenosis is rarely present as long assome intraluminal contrast enhancement can be de-tected within the segment in question. Assessment tendsto be more accurate in the absence of motion artifactsand calcification. Calcification especially may lead tooverestimation of luminal narrowing. The question ofwhether quantitative measurements are helpful in in-creasing diagnostic accuracy has yet to be answered.

FINDINGS

Coronary Anatomy

A thorough understanding of the coronary artery anat-omy is a prerequisite for the correct diagnostic interpreta-

Figure 2. Axial CT image obtained at level of left main coro-nary artery (LM) ostium. Left main coronary artery arises fromleft sinus of Valsalva and runs posterior to pulmonary artery.Left main coronary artery divides into LAD (LAD) and left cir-cumflex coronary artery (LCX). In this image, trifurcation ispresent. Ramus intermedius (RI ) arises from left main coronaryartery. RVOT = right ventricular outflow tract.

Card

iac Ap

plicatio

ns fo

r Mu

lti–Detecto

r Row

CT

63

tion of CT coronary angiographic findings. Major coro-nary arteries are well delineated and easy to evaluate ongood-quality CT coronary angiograms obtained with thecurrent CT technology (16-section multi–detector rowCT, submillimeter collimation, temporal resolution of210 msec). For a better understanding of the coronaryanatomic structures, the appearance of the major coro-nary arteries in contiguous axial CT images is described.

Typically, the American Heart Association’s 16-seg-ment model of the coronary arteries is used to describethe anatomic structures. Two coronary arteries originatefrom the aorta. The left main coronary artery arisesfrom the left sinus of Valsalva and courses to the leftand posterior to the main pulmonary artery. On the leftside, the left main coronary artery divides into the leftanterior descending coronary artery (LAD) and left cir-cumflex coronary artery (Fig 2). The LAD runs anteri-orly in the anterior interventricular groove and gives offdiagonal and septal branches (Fig 3). The LAD can befollowed to the apex of the heart. The left circumflexcoronary artery runs in the left atrioventricular grooveclose to the great cardiac vein (Fig 3). The left circum-flex coronary artery gives off obtuse marginal branches,which supply the lateral wall of the left ventricle.

The right coronary artery originates from the rightsinus of Valsalva. The ostium of the right coronary ar-tery is typically caudal to the origin of the left maincoronary artery in the axial sections. The right coro-nary artery has a short horizontal proximal segmentthat runs anteriorly and to the right. More distally, theright coronary artery courses caudally in the rightatrioventricular groove. The mid segment of the right

coronary artery is commonly shown in cross sectionin axial images (Fig 3). The distal segment of the rightcoronary artery is located in the atrioventriculargroove on the inferior surface of the heart and reachesthe posterior crux of the heart.

The posterior descending coronary artery arisesfrom the right coronary artery in 70% of people (ie,right dominance) and runs in the posterior interven-tricular groove parallel to the middle cardiac vein. In10% of people, the left circumflex coronary arteryreaches the crux of the heart and continues as the pos-terior descending coronary artery (ie, left dominance).In 20% of people, the right coronary artery gives riseto the posterior descending coronary artery, but theleft circumflex coronary artery gives rise to branches,which supply the inferior wall of the left ventricle (ie,balanced system or codominance).

For identification of coronary artery calcification,coronary anatomy is more difficult because the lumenof the coronary vessels is not enhanced. Quantifica-tion of coronary artery calcium is performed for theleft main coronary artery, the LAD, the left circumflexcoronary artery, the right coronary artery, and the pos-terior descending coronary artery.

There are common pitfalls in the evaluation of CTcoronary angiographic findings. The LAD may extendabove the level of the left main artery ostium. Coronaryveins may be confused with coronary arteries or mayoverlay coronary arteries; specifically, the great cardiacvein frequently overlaps the proximal LAD, ramus in-termedius, and proximal left circumflex coronary ar-tery. However, scrolling through the axial sections ofthe data set, where vascular structures can typically befollowed, allows the separation of veins and arteries. Inaddition, specifically for coronary calcium scoring, arti-facts (eg, beam hardening, pacemaker leads, intracoro-nary stents, valve replacements, bypass clips) may befalsely scored. Also, mitral valve annulus or leaflet calci-fications may be confused with midsegment left cir-cumflex coronary artery calcifications. The diagnosticevaluation of coronary CT angiographic findings maybe further enhanced with postprocessing of the originaldata set by using techniques discussed previously (eg,MIPS, volume rendering).

Pathologic Findings

Typical pathologic findings in patients referred for car-diac CT in the emergency department include disease ofthe coronary arteries, myocardium, and paracardiacstructures. The diagnostic implications of such findingsare discussed in the following paragraphs. The cause ofchest pain includes more than 50 different diagnosticpossibilities. The most important differential diagnosesinclude aortic dissection and pulmonary embolism,which are discussed extensively elsewhere in this sylla-bus. For a complete overview of differential diagnoses,we refer the reader to Gibbons et al (11).

Figure 3. Axial CT image. Mid and distal segments of LAD(LAD) are located in interventricular groove. Left circumflex cor-onary artery (LCX ) runs in left atrioventricular groove close togreat cardiac vein (GCV ). RCA = right coronary artery.

Ho

ffm

ann

et

al

64

Cardiac CT imaging is unlikely to be performed inpatients with known myocardial infarction, ongoingangina, or hemodynamic instability. However, pa-tients with a typical clinical manifestation, nonspecificECG changes (ST-segment depression < 1 mm, T-wave

inversion), and negative initial serum cardiac markersor patients with unstable angina may profit from de-tection or exclusion of significant coronary artery ste-nosis (Fig 4). The nature of the occlusion may beidentified in some patients (eg, thrombotic occlusion

Figure 4. Images of a patientwith unstable angina. (a) Five-millimeter thin-slab MIP multi–detector row CT image showsstenosis (arrow) in right coro-nary artery. (b) Invasive selec-tive coronary angiographicimage confirms presence ofsignificant stenosis (arrow) atidentical location.

Figure 5. Images of a patient with acute myo-cardial infarction who underwent successfulthrombolysis. (a) Coronary angiographic imageshows occlusion (arrow) of LAD (LAD) just distalto second diagonal branch (2DIAG). 1DIAG =first diagonal branch. (b) Multi–detector row CTimage demonstrates perfusion deficit (area be-tween white arrows) in same region supplied bydistal LAD, in anteroseptal segment of mid andapical portions of left ventricular myocardium,representing area of acute myocardial infarction.Note large thrombus (arrowhead) within left ven-tricular cavity. LA = left atrium, LV = left ventri-cle, RV = right ventricle. (c) Volume-renderedmulti–detector row CT image obtained after suc-cessful thrombolysis during invasive coronaryangiography demonstrates patent distal LAD(arrow) after second diagonal branch (2DIAG).LAD = left anterior descending coronary artery,LCX = left circumflex artery, 1DIAG = first diago-nal branch.

Card

iac Ap

plicatio

ns fo

r Mu

lti–Detecto

r Row

CT

65

Figure 7. Images of a patientobtained after coronary arterybypass surgery. (a) Volume-ren-dered multi–detector row CT im-age shows patent left internalmammary artery graft to LAD(arrows) and saphenous veingraft to first obtuse marginalbranch (arrowheads). (b) Vol-ume-rendered multi–detector rowCT image demonstrates patentsaphenous vein graft (arrows) toposterior descending coronaryartery.

Figure 6. Multi–detector rowCT images of acute thromboticocclusion of left circumflex ar-tery in a patient with recent myo-cardial infarction. (a) Curvedmultiplanar reformation image ofleft circumflex artery showsproximal stenosis (arrowhead)and midthrombotic occlusion (ar-row). (b) Cross-sectional multi-planar reformation image of leftcircumflex artery shows typicalappearance of thrombotic occlu-sion (arrow).

of long segments of the artery) (Fig 5). In patientswith acute myocardial infarction who receive systemicthrombolysis without percutaneous intervention, car-diac multi–detector row CT may be used to demon-strate the success of thrombolysis (Fig 6).

Because the acquired CT data permit the assessmentof left ventricular function, the assessment of globaland regional wall motion abnormities, myocardialthinning consistent with scar tissue, and previousmyocardial infarction may be possible. CT may alsobe used to depict thrombi (Fig 6) typically found inthe presence of left ventricular aneurysms and im-paired left ventricular function. The findings in somecases suggest that CT may be able to be used to depictperfusion deficits in acute myocardial infarction. Inpatients with coronary artery bypass grafts, graft pa-tency can be assessed (Fig 7).

In patients who have a history of chest pain or atypi-cal chest pain, nonobstructive calcified and noncal-cified coronary atherosclerotic plaques may be de-tected (Figs 8, 9). In addition, although the condition

Figure 8. Image of a patient with atypical chest pain. Thin-slab MIP reconstruction demonstrates two nonstenotic calcifiedand noncalcified plaques (arrows) of mid right coronary artery.

Ho

ffm

ann

et

al

66

Figure 9. Images demonstrating positive remodeling in left main coronary artery of a patient with a history of chest pain. (a) Curvedreconstruction of left main coronary artery and LAD demonstrates positive remodeling with small calcified and noncalcified plaque(arrow) in left main coronary artery. (b) Cross-sectional image at level of the left main coronary artery shows mixed plaque (arrow)with calcified and noncalcified components.

Figure 10. Images showinganomalous right coronary arteryof a patient with syncope andexertional chest pain. (a) Thin-slab MIP shows anomalous ori-gin of right coronary artery (ar-row) from left sinus of Valsalva,with narrowing of proximal rightcoronary artery segment when ittravels between main pulmonaryartery (MPA) and ascendingaorta (AO). Note normal originof left main coronary artery (ar-rowhead). (b) Volume-renderedimage also demonstrates anom-alous origin of right coronary ar-tery (arrow) from left sinus ofValsalva, adjacent to origin ofleft main coronary artery (arrow-head).

is rare, anomalous origin of a coronary artery may bedetected (Fig 10).

DIAGNOSTIC VALUE OF MULTI–DETECTOR ROW CT

Detection of Significant Coronary Artery Stenosisin Patients with Stable Angina

Although the examples in the preceding paragraphssuggest that multi–detector row CT may be used toimprove the diagnosis of acute coronary syndrome inpatients with acute chest pain, no systematic analysis

of the diagnostic value of cardiac multi–detector rowCT in the emergency department has been reported.We summarize here the results of available studiesthat have been conducted in patients with stable an-gina. Considerable selection bias is prevalent through-out the individual case series because only patientswho were highly suspected of having significant coro-nary artery disease were included.

Early studies with electron-beam CT and early ver-sions of multi–detector row CT scanners (equippedwith four detectors and a temporal resolution of 250–

Card

iac Ap

plicatio

ns fo

r Mu

lti–Detecto

r Row

CT

67

330 msec) showed that cardiac CT could be used todetect significant coronary stenosis with high sensitiv-ity and excellent specificity: For electron-beam CT, thesensitivity was 82% ± 6.4, and the specificity 87% ±0.6, and for multi–detector row CT, the values were81% ± 7.2 and 91% ± 0.9, respectively, when com-pared with selective conventional coronary angiogra-phy in segments that could be evaluated. However,the clinical applicability of these methods was limitedby incomplete CT evaluation of coronary anatomicstructures: Roughly one-third of all coronary arterysegments could not be evaluated because of stair-stepand motion artifacts and the presence of calcification.If all coronary segments were included, diagnostic ac-curacy substantially decreased for electron-beam CTand for multi–detector row CT (sensitivity, 68% ± 4.3and 63% ± 2.9, respectively; specificity, 68% ± 0.2 and71% ± 0.2, respectively) (12−18).

Initial reports of 16-section technology are promis-ing. Results show improved diagnostic accuracy forthe depiction of significant stenosis in assessable seg-ments (92% sensitivity, 93% specificity) comparedwith elective conventional coronary angiography. Im-proved image quality achieved with better spatial andtemporal resolution led to a substantial decrease ofthe number of segments that could not be evaluatedin coronary vessels more than 1.5 mm in diameter(12%). Subsequently, sensitivity and specificity withthe inclusion of all segments increased considerably(84% and 85%, respectively) (19,20). Results from arecent meta-analysis are summarized in Table 1.

In addition to the detection of coronary artery ste-noses and occlusions, CT imaging provides other in-formation of potential value in patients with acutechest pain. Global and regional left ventricular func-tion can be assessed with high accuracy. The depiction

of myocardial perfusion deficits has been demon-strated, and the results of preclinical studies have il-lustrated the potential to assess microvascular func-tion.

Comparison with Magnetic Resonance Imaging

Magnetic resonance (MR) imaging is being inten-sively explored because of its potential for coronaryvessel depiction (21−24). In spite of impressive ad-vances, MR still requires temporally prolonged acqui-sitions and data averaging during several (as many as40) heart beats to generate one image, which substan-tially limits image quality. In addition, a sufficient sig-nal-to-noise ratio relies on relatively thick (1.5-mm)sections that importantly limit spatial resolution. In arecent multicenter study of 109 patients, MR imagingdemonstrated moderate accuracy (72%) for the detec-tion of significant coronary artery stenosis (25). Inone study, investigators assessed the value of MR im-aging in 161 patients with chest pain who had anondiagnostic ECG for acute myocardial infarctionand found a significant benefit compared with theusual clinical care and diagnostic tests; the benefit wasderived predominantly from accurate assessment ofleft ventricular function at rest (26).

Despite the natural advantage of MR for soft-tissueimaging, artifacts caused by cardiac and respiratorymotion and the relatively low spatial resolution re-main challenges for coronary MR imaging. More im-portant, long imaging times limit its applicability inthe setting of evaluating patients in the emergency de-partment. Consequently, coronary MR imaging israrely available in emergency departments and cur-rently would be less feasible to incorporate into therapid evaluation of patients with chest pain, com-pared with multi–detector row CT.

Table 1Pooled Measures of Accuracy from a Meta-analysis of the Current Literature Regarding CT for the Detection ofSignificant Coronary Artery Stenosis

Multi–Detector Row CT

Electron-beam CT 4 Detector Rows 16 Detector Rows

Simple Method Mean LL 95% UL 95% Mean LL 95% UL 95% Mean LL 95% UL 95%

Assessable segments onlyWeighted mean sensitivity (%) 82 79 85 84 81 88 92 87 97Weighted mean specificity (%) 87 84 89 93 91 96 93 89 97

All segmentsWeighted mean sensitivity (%) 69 66 72 67 63 71 83 76 89Weighted mean specificity (%) 67 64 71 77 74 81 85 79 91

Per patientWeighted mean sensitivity (%) 81 78 84 86 83 89 91 87 96Weighted mean specificity (%) 54 51 57 80 77 84 82 76 89

Note.—LL = lower limit, UL = upper limit.

Ho

ffm

ann

et

al

68

Coronary Artery Calcification in Patients withAcute Chest Pain

The only experience with cardiac CT in patients withacute chest pain comes from the results of a few stud-ies in which electron-beam CT depiction of coronaryartery calcification was used to predict the likelihoodof acute coronary syndrome in patients with acutechest pain (Table 2) (27−29). In those studies, investi-gators demonstrated that the absence of coronary cal-cifications had a high negative predictive value foracute coronary syndrome. Although these results arepromising, imaging of coronary calcification is notroutinely used in the diagnostic work-up of patientswith acute chest pain. Indeed, the role of coronary cal-cification is controversially discussed. Among indi-viduals dying of sudden cardiac death, calcium wasfound in only 50% of the coronary artery “culprit” le-sions. In addition, the absence of coronary calcifica-tion does not imply the absence of any coronary ath-erosclerotic plaque, especially in young patients.Moreover, noncalcified plaque may be more prone torupture and cause symptoms and events than calcifiedplaque because noncalcified plaque consists of a lipidcore and a thin “unstable” fibrous cap.

In summary, diagnostic tools that permit the reli-able and rapid triage of patients with acute coronarysyndrome are urgently needed. Imaging may play aprominent role in that work-up if it provides accurateinformation in an economically reasonable way. CTimaging with multi–detector row CT (16 sections ormore) has matured to a stage that permits reliable de-piction of the coronary arteries after intravenous injec-tion of a contrast agent. Currently available data sug-gest a high negative predictive value of multi–detectorCT in the assessment of hemodynamically relevantcoronary artery lesions in the absence of calcificationsand motion artifacts. Although no study has specifi-cally investigated the usefulness of cardiac CT in thesetting of acute chest pain, the ability of multi–detec-tor row CT to provide additional information on cor-onary atherosclerotic plaque burden and left ventricu-lar function makes cardiac CT likely to be beneficiallyintegrated into the work-up of patients with acutechest pain. However, dedicated trials have not been

completed, and we must await their results before rec-ommendations for use can be made.

References1. Pope JH, Aufderheide TP, Ruthazer R, et al. Missed diag-

noses of acute cardiac ischemia in the emergency depart-ment. N Engl J Med 2000; 342:1163–1170.

2. Lee TH, Rouan GW, Weisberg MC, et al. Clinical character-istics and natural history of patients with acute myocardialinfarction sent home from the emergency room. Am JCardiol 1987; 60:219–224.

3. Lee TH, Goldman L. Evaluation of the patient with acutechest pain. N Engl J Med 2000; 342:1187–1195.

4. Goldman L, Cook EF, Johnson PA, Brand DA, Rouan GW,Lee TH. Prediction of the need for intensive care in patientswho come to the emergency department with acute chestpain. N Engl J Med 1996; 334:1498–1504.

5. Fleischmann KE, Hunink MG, Kuntz KM, Douglas PS. Exer-cise echocardiography or exercise SPECT imaging? a meta–analysis of diagnostic test performance. J Nucl Cardiol2002; 9:133–134.

6. Klocke FJ, Baird MG, Lorell BH, et al. ACC/AHA/ASNCguidelines for the clinical use of cardiac radionuclide imag-ing: executive summary—a report of the American Collegeof Cardiology/American Heart Association Task Force onPractice Guidelines (ACC/AHA/ASNC Committee to Revisethe 1995 Guidelines for the Clinical Use of Cardiac Radio-nuclide Imaging). Circulation 2003; 108:1404–1418.

7. Cheitlin MD, Armstrong WF, Aurigemma GP, et al. ACC/AHA/ASE 2003 guideline update for the clinical applicationof echocardiography: summary article—a report of theAmerican College of Cardiology/American Heart Associa-tion Task Force on Practice Guidelines (ACC/AHA/ASECommittee to Update the 1997 Guidelines for the ClinicalApplication of Echocardiography). Circulation 2003; 108:1146–1162.

8. Achenbach S, Moshage W, Ropers D, Bachmann K.Curved multiplanar reconstructions for the evaluation ofcontrast-enhanced electron-beam CT of the coronary arter-ies. AJR Am J Roentgenol 1998; 170:895–899.

9. Achenbach S, Ropers D, Regenfus M, et al. Diagnosticvalue of 2- and 3-dimensional image reconstruction tech-niques for the detection of coronary artery stenoses by con-trast-enhanced electron-beam CT (abstr). J Am Coll Cardiol2000; 35(suppl A):416.

10. Ropers D, Moshage W, Daniel WG, Jessl J, Gottwik M,Achenbach S. Visualization of coronary artery anomaliesand their anatomic course by contrast-enhanced electronbeam tomography and three-dimensional reconstruction.Am J Cardiol 2001; 87:193–197.

11. Gibbons RJ, Chatterjee K, Daley J, et al. ACC/AHA/ACP-ASIM guidelines for the management of patients with

Table 2Predictive Value of the Presence or Absence of Coronary Artery Calcification for Coronary Events in Patients with AcuteChest Pain and Uncertain Myocardial Infarction

Positive NegativeNo. of Predictive Predictive

Reference Year Modality Patients Sensitivity Specificity Value Value

Laudon et al (27) 1999 Electron-beam CT 105 1.00 NA NA 1.00McLaughlin et al (28) 1999 Electron-beam CT 134 0.88 0.37 0.08 0.98Georgiou et al (29) 2001 Electron-beam CT 192 1.00 0.47 0.26 1.00

Note.—NA = not available.

Card

iac Ap

plicatio

ns fo

r Mu

lti–Detecto

r Row

CT

69

chronic stable angina: a report of the American College ofCardiology/American Heart Association Task Force onPractice Guidelines (Committee on Management of Patientswith Chronic Stable Angina). J Am Coll Cardiol 1999; 33:2092–2197. [Errata: J Am Coll Cardiol 1999; 34:314; J AmColl Cardiol 2001; 38:296.]

12. Achenbach S, Giesler T, Ropers D, et al. Detection of coro-nary artery stenoses by contrast-enhanced, retrospectivelyelectrocardiographically-gated, multislice spiral computedtomography. Circulation 2001; 103:2535–2538.

13. Achenbach S, Moshage W, Ropers D, Nossen J, DanielWG. Value of electron-beam computed tomography for thenoninvasive detection of high-grade coronary-artery steno-ses and occlusions. N Engl J Med 1998; 339:1964–1971.

14. Becker CR, Knez A, Leber A, et al. Detection of coronaryartery stenoses with multislice helical CT angiography. JComput Assist Tomogr 2002; 26:750–755.

15. Flohr T, Ohnesorge B. Heart rate adaptive optimization ofspatial and temporal resolution for electrocardiogram-gatedmultislice spiral CT of the heart. J Comput Assist Tomogr2001; 25:907–923.

16. Kopp AF, Schroeder S, Kuettner A, et al. Non-invasive cor-onary angiography with high resolution multidetector-rowcomputed tomography: results in 102 patients. Eur Heart J2002; 23:1714–1725.

17. Kopp AF, Schroeder S, Kuettner A, et al. Coronary arteries:retrospectively ECG-gated multi–detector row CT angiogra-phy with selective optimization of the image reconstructionwindow. Radiology 2001; 221:683–688.

18. Ohnesorge B, Flohr T, Becker C, et al. Cardiac imaging bymeans of electrocardiographically gated multisection spiralCT: initial experience. Radiology 2000; 217:564–571.

19. Ropers D, Baum U, Pohle K, et al. Detection of coronary ar-tery stenoses with thin-slice multi–detector row spiral com-puted tomography and multiplanar reconstruction. Circula-tion 2003; 107:664–666.

20. Nieman K, Cademartiri F, Lemos PA, Raaijmakers R,Pattynama PM, de Feyter PJ. Reliable noninvasive coro-

nary angiography with fast submillimeter multislice spiralcomputed tomography. Circulation 2002; 106:2051–2054.

21. Shinnar M, Fallon JT, Wehrli S, et al. The diagnostic accuracyof ex vivo MRI for human atherosclerotic plaque characteriza-tion. Arterioscler Thromb Vasc Biol 1999; 19:2756–2761.

22. Fayad ZA, Fuster V, Fallon JT, et al. Noninvasive in vivohuman coronary artery lumen and wall imaging using black-blood magnetic resonance imaging. Circulation 2000; 102:506–510.

23. Botnar RM, Stuber M, Kissinger KV, Kim WY, Spuentrup E,Manning WJ. Noninvasive coronary vessel wall and plaqueimaging with magnetic resonance imaging. Circulation 2000;102:2582–2587.

24. Kim WY, Stuber M, Bornert P, Kissinger KV, Manning WJ,Botnar RM. Three-dimensional black-blood cardiac mag-netic resonance coronary vessel wall imaging detects posi-tive arterial remodeling in patients with nonsignificant coro-nary artery disease. Circulation 2002; 106:296–299.

25. Kim WY, Danias PG, Stuber M, et al. Coronary magneticresonance angiography for the detection of coronary steno-ses. N Engl J Med 2001; 345:1863–1869.

26. Kwong RY, Schussheim AE, Rekhraj S, et al. Detectingacute coronary syndrome in the emergency departmentwith cardiac magnetic resonance imaging. Circulation 2003;107:531–537.

27. Laudon DA, Vukov LF, Breen JF, Rumberger JA, WollanPC, Sheedy PF II. Use of electron-beam computed tomog-raphy in the evaluation of chest pain patients in the emer-gency department. Ann Emerg Med 1999; 33:15–21.

28. McLaughlin VV, Balogh T, Rich S. Utility of electron beamcomputed tomography to stratify patients presenting to theemergency room with chest pain. Am J Cardiol 1999; 84:327–328, A8.

29. Georgiou D, Budoff MJ, Kaufer E, Kennedy JM, Lu B,Brundage BH. Screening patients with chest pain in theemergency department using electron beam tomography: afollow-up study. J Am Coll Cardiol 2001; 38:105–110.

NO

TES

71

Imaging of BluntChest Trauma1

Trauma is the leading cause of death in Americans younger than 45 years of age (1).Blunt chest trauma is the second most common cause of death in trauma patients, fol-lowing injury to the central nervous system. This chapter will discuss the imaging find-ings of chest trauma, organized by specific organ or structure, including lung paren-chyma, pleura, airway, and chest wall. Injury to the aorta and great vessels will be dis-cussed elsewhere.

MECHANISM OF INJURY IN BLUNT CHEST TRAUMA

Three main mechanisms of injury are associated with blunt chest trauma: (a) rapid ac-celeration and/or deceleration, (b) direct impact, and (c) thoracic compression (2,3).

Acceleration/deceleration injuries are often caused by a motor vehicle accident or a fallfrom a height (4), leading to shearing forces on tissues, organs, and blood vessels thatcan result in tissue disruption. The most lethal shearing injury is aortic rupture, whichcan be fatal within seconds (3).

Direct-impact injuries result from a motor vehicle accident, a fall, or a direct blowfrom a moving object, leading to localized chest wall injury, such as fractures of theribs, sternum, and scapulae and soft-tissue hematoma. Direct-impact forces can alsoinjure the lungs and heart because kinetic energy is transmitted through the chest wallinto the deeper tissues. Secondary penetrating injuries may also be seen.

Compression injuries often occur in the setting of rapid deceleration as tissues strikea fixed object such as the chest wall or spine, leading to organ rupture, contusion, orhemorrhage (5).

IMAGING FOR BLUNT CHEST TRAUMA

Supine anteroposterior chest radiography is the initial, as well as the most common,imaging examination used for evaluating the condition of a patient who is suspectedof having blunt chest injury. Many common injuries are identified with the chest radio-graph alone. However, in severely injured patients, the ideal upright full-inspiratoryposteroanterior chest radiograph cannot be obtained, and suboptimal supine radio-graphs, often with poor positioning, poor inspiration, and underlying backboard andoverlying monitoring equipment, are generally the rule, rather than the exception.These factors can easily hinder depiction of injuries.

Nisa Thoongsuwan, MD, Jeffrey P. Kanne, MD,and Eric J. Stern, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 71–79.

1From the Department of Radiology, Harborview Medical Center, 325 Ninth Ave, Box 359728, Seattle, WA 98104-2499(e-mail: nisa@u.washington.edu).

Tho

on

gsu

wan

et a

l

72

Computed tomography (CT) of the chest, particu-larly in the era of multi–detector row CT technology,has become a common examination for imaging thetrauma patient with known or suspected thoracic in-jury because CT scanners are available in almost alltrauma centers and because scan times have mark-edly decreased. The clear advantage of chest CT overradiography is the ability of chest CT to depict oc-cult injuries that may not be evident on a supinechest radiograph, such as mediastinal hematoma,pneumothorax, or hemopericardium (6).

Transesophageal echocardiography may have a rolein evaluating patients with blunt chest trauma. De-spite its dependence on the skill of the operator,transesophageal echocardiography has been found tobe particularly useful in assessing the descending tho-racic aorta for injury, as well as cardiac structure andfunction in selected cases (7).

These imaging studies are an essential component ofevaluating the trauma patient after assessing and stabi-lizing the airway and respiratory and cardiac function.It is essential to identify life-threatening conditions onthe chest radiograph, such as pneumothorax, hemo-thorax, abnormal mediastinum (possibly suggestive ofaortic or other great vessel injury), and thoracic spinefracture, as well as malpositioned life-support devices.The technical limitations of a chest radiograph shouldbe declared when it is difficult or impossible to ex-clude a life-threatening injury, and alternative imagingstudies should be suggested (8).

LUNG PARENCHYMA

Pulmonary Contusion

Pulmonary contusion (Fig 1) is a term that describesinterstitial and alveolar injury without substantiallaceration. Pulmonary contusion is the most com-mon pulmonary injury resulting from blunt chesttrauma (9). Pathologically, both hemorrhage andtransudate fill the alveolar interstitium followingloss of alveolar capillary integrity without accompa-nying major parenchymal disruption (10). Themechanism for this injury involves local compres-sive and recoil forces within the lung (9,11–14).

Although the radiographic findings of pulmonarycontusion are nonspecific, ranging from irregularconfluent or discrete nodular opacities to largeopacities, the time course of the development andevolution of the opacity is the key to identifying thisinjury (12). Pulmonary contusions typically appearwithin hours of injury and clear within 7 days (12).Moreover, the pulmonary opacities found in pulmo-nary contusion occur in a nonanatomic distribution,in contrast to opacities resulting from pneumonia oraspiration (15), because the energy absorbed has norespect for bronchovascular segmental anatomicstructures.

CT is highly sensitive and is more specific thanchest radiography for identifying pulmonary contu-sion (16). Pulmonary contusion appears as an ill-defined area of peripheral consolidation at CT. Al-though chest radiographs are useful, they may notdepict a contusion early in its course (1) becausethe contusion may be obscured by comorbidities.

Pulmonary Laceration

Pulmonary laceration is a tear in the lung paren-chyma. The mechanism of injury centers on eithershearing forces from blunt trauma or direct punc-ture from a penetrating injury (9,12–14). As a result

Figure 1. Supine chest radiograph of a 12-year-old boy in-jured in high-speed motor vehicle accident shows peripheral pa-renchymal opacity of left lung, compatible with pulmonary con-tusion. Also note soft-tissue swelling of left chest wall (arrows).

Figure 2. CT scan of a 36-year-old woman injured in high-speed motor vehicle accident shows round air-containing lesionin right lung, consistent with type 1 pulmonary laceration.

Imag

ing

of Blu

nt C

hest Trau

ma

73

of the recoil properties of the adjacent lung, the ini-tial linear parenchymal tear rapidly becomes anovoid or round space (9,12). When the lacerationfills with blood, it can be called a hematoma. If thespace fills with air, it can be referred to as a trau-matic pneumatocele. Frequently, both blood andair are present; therefore we prefer to use the termpulmonary laceration to cover all scenarios.

The typical radiographic appearance of pulmo-nary laceration is a round lesion containing eitherair or both air and fluid. Pulmonary lacerationscommonly are seen as isolated lesions but may bemultiple. Most pulmonary lacerations are 2–5 cm indiameter, but they can occasionally be extremelylarge, as much as 14 cm in diameter. Pulmonarylacerations are present at the time of injury but may

be obscured by surrounding pulmonary contusion,hemothorax, and pneumothorax (9,17).

Pulmonary lacerations can be categorized intofour types on the basis of the mechanism of injury,as described by Wagner et al (12). A type 1 pulmo-nary laceration (Fig 2) is typically a large (2–8-cm)cavity filled with a variable amount of air and bloodoccurring deep within the pulmonary parenchyma.Type 1 pulmonary laceration results from suddencompression of the pliable chest wall against theclosed glottis, which causes the air-containing pa-renchyma to rupture.

Type 2 pulmonary laceration (Fig 3) also resultsfrom rapid compression of the chest wall. In thiscase, however, lung injury occurs from shearingforces as the lung is squeezed over the vertebralbodies. The resulting laceration typically occurs inthe paraspinal lung parenchyma. These lacerationscan be long and tubular, rather than spherical.

A type 3 pulmonary laceration (Fig 4) typicallyappears as a small peripheral area of low attenua-tion intimately associated with an adjacent rib frac-ture and is caused by puncture with a fractured rib.Type 3 pulmonary lacerations are often multiple.

Type 4 pulmonary laceration is the result of a pre-viously formed, firm pleuropulmonary adhesioncausing the lung to tear when the overlying chestwall is violently compressed inward or is fractured.This type of pulmonary laceration is almost alwaysand only identified at surgery or autopsy.

The findings from one study showed that themost common type of pulmonary laceration is type1, followed by type 3 and type 2. Type 4 pulmonarylacerations were rare (12).

Figure 3. (a) Supine chest radiograph of a 15-year-old girl involved in high-speed motor vehicle accident shows elliptical air-fluidlesion at right cardiophrenic sulcus (arrow), representing type 2 pulmonary laceration. (b) Transverse CT scan better defines pulmo-nary laceration (arrow).

Figure 4. Transverse CT scan of a 25-year-old man injured inhigh-speed motor vehicle accident shows round air-containinglesion (white arrow) in right lung adjacent to fractured rib (blackarrow), consistent with type 3 pulmonary laceration.

Tho

on

gsu

wan

et a

l

74

PLEURA

Pneumothorax

Pneumothorax (Fig 5) is a collection of gas in thepleural space. Pneumothorax occurs in 15%–38%of the patients with blunt chest trauma (18,19). Be-cause pneumothorax is usually associated with ribfracture, the mechanism of injury is generally a di-rect puncture of the visceral pleura, with subsequentair leakage into the pleural space (20).

Although the clinical importance of the pneumo-thorax is not dependent on its size, but rather onthe underlying cardiopulmonary function of the pa-tient, almost all pneumothoraces in victims oftrauma should be considered clinically important,given that pneumothoraces can rapidly become lifethreatening when patients receive general anesthesiaor are treated with positive pressure mechanicalventilation (21,22).

As previously mentioned, in the setting of trauma,the chest radiograph is usually obtained with the pa-tient supine. When the patient is supine, free intra-pleural gas preferentially collects in the nondepen-dent anteromedial and inferior aspect of the pleuralspace. Consequently, the radiographic findings ofpneumothorax on supine radiographs are differentfrom those seen on upright radiographs. Radio-graphic features of pneumothorax on the supine ra-diograph include the deep sulcus sign (prominenceof the costophrenic angle), basilar hyperlucency, un-usual sharp delineation of the mediastinal or cardiaccontour, and clear depiction of the apical pericardialfat pad (8,14,23). When an upright chest radiographcannot be obtained and in the appropriate clinicalsetting, lateral decubitus or cross-table lateral radio-graphs may demonstrate a pneumothorax.

CT is the most accurate method for detecting pneu-mothorax (24). The advantages of CT over radiogra-phy are (a) CT is performed with the patient supine,and (b) CT can be used to distinguish pneumothoraxfrom gas in the overlying soft tissues (8). Because thesmallest pneumothorax can develop into a life-threat-ening tension pneumothorax, chest CT should be con-sidered in patients with no evidence of pneumothoraxon the supine radiograph who are at risk for pneumo-thorax and who will receive positive pressure ventila-tion (8,9,14). Some trauma centers also routinely ob-tain CT images through the lung bases as a part of anabdominal trauma CT protocol, potentially identify-ing radiographically occult pneumothoraces (24).When carefully sought, pneumothoraces can also bedepicted on cervical or thoracic spine CT scans.

Tension pneumothorax is one of the most com-mon life-threatening intrathoracic injures caused byblunt trauma (25). The diagnosis in most cases ismade from clinical signs and symptoms, which in-clude dyspnea, hypoperfusion, jugular venous disten-tion, diminished breath sounds on the affected side,hyperresonance to percussion, and tracheal shift tothe unaffected side (1). The radiographic findings re-sulting from tension pneumothorax include (a) in-creased lucency of the affected hemithorax withcontralateral displacement of the mediastinum andtrachea and (b) flattening or even inversion of theipsilateral hemidiaphragm (20).

Hemothorax

Hemothorax is a collection of blood within thepleural space. Bleeding from low-pressure vesselsmay subside spontaneously or following placementof a pleural drain (8). However, massive hemotho-

Figure 5. Supine anteroposterior chest radiograph of a 54-year-old man who fell 20 ft (6 m) shows increased lucency ofright hemithorax and deep sulcus sign (arrows), indicative ofright pneumothorax.

Figure 6. Nonenhanced transverse CT scan of an 83-year-oldwoman who fell shows bilateral pleural collections. The fluid-fluid level in right pleural space (arrow) is referred to as hemat-ocrit sign and is indicative of hemothorax.

Imag

ing

of Blu

nt C

hest Trau

ma

75

rax is a life-threatening condition because of poten-tial mass effect on the heart and great vessels fromthe accumulated blood, acute hypovolemic shock,and hypoxia from lung collapse (2).

The findings of hemothorax on the supine chest ra-diograph are often indirect. They include (a) diffuselyincreased opacity through the affected hemithorax,(b) a homogeneous crescent-shaped opacity inter-posed between the inner margin of the ribs and thelung, or (c) an apical cap (a crescent opacity over thelung apices on the supine radiograph) (8).

At CT, particularly in the acute setting, bloodproducts in the pleural space may have increased at-tenuation, and when active bleeding is present, lay-ering of fluids with different attenuation may occur.This layering is referred to as a hematocrit layer orthe hematocrit sign (Fig 6).

AIRWAY

Tracheobronchial Laceration

Among patients sustaining blunt chest injury, theoccurrence of tracheobronchial laceration is rare. Inthis group, tracheal rupture accounts for approxi-mately 15%–27% of all tracheobronchial lacera-tions and is associated with higher overall morbid-ity and mortality (26–28). The diagnosis of trachealrupture may be delayed because of its rarity and itsoften nonspecific clinical and radiographic mani-festations. In addition, other more common associ-ated injuries, such as the rupture of the great ves-sels, may mask underlying tracheobronchial lacera-tion (27,29).

The injuries to the tracheobronchial tree can bedivided into intrathoracic and extrathoracic types.We will discuss only the intrathoracic type. Mecha-

nisms of intrathoracic tracheobronchial lacerationinclude a sudden increase in intra-airway pressureagainst a closed glottis at the time of impact. Thisincrease can cause a tear across the lower tracheo-carinal junction from anteroposterior chest com-pression forcing the lungs apart laterally. Othermechanisms include hyperextension of the neck, di-rect crush injury of the trachea between the sternumand thoracic spine, and sudden and rapid decelera-tion with shearing force applied to the relativelyfixed cricoid cartilage and carina (26,28,30).

The most common radiographic manifestationsof tracheobronchial laceration are pneumomedi-astinum and pneumothorax, occurring in approxi-mately 70% of patients (31). The “fallen lung” sign,while diagnostic, rarely occurs and represents com-plete disruption of all anchoring attachment of thelung to the hilum. The transected lung falls againstthe posterolateral chest wall or hemidiaphragm(32–34), and there is a dramatic hydropneumotho-rax. This sign is more readily apparent at CT thanon the chest radiograph (35) (Fig 7).

CT of the chest with an appropriate window set-ting can frequently show the exact site of a tear,manifesting as a focal defect in or a circumferentialabsence of the tracheal or bronchial wall, a centralairway wall contour deformity, abnormal commu-nication of the central airway with other mediasti-nal structures (36–39), overdistention of the endo-tracheal tube balloon, herniation of the deformedendotracheal balloon beyond the trachea, or extra-luminal location of the endotracheal tube (40). In-direct signs, such as deep cervical emphysema andpneumomediastinum, should raise suspicion fortracheobronchial injury in the appropriate clinicalsetting (28,41–43).

Figure 7. Images of an 18-year-old man injured in motor vehicle accident. (a) Supine radiograph shows right pneumomediastinumand right pneumothorax. (b) Transverse CT scan demonstrates tracheal laceration (arrow). Note the collapsed lung (∗) has fallenaway from the hilum inferiorly and laterally, which is also called the fallen lung sign.

Tho

on

gsu

wan

et a

l

76

Diaphragmatic Injury

Diaphragmatic injury (Fig 8) occurs in as many as8% of the patients with blunt trauma injuries (44), oc-curring most often in young men injured in motor ve-hicle accidents (11,45). Diaphragmatic injury can be adiagnostic challenge because of occasional subtle imag-ing findings or the presence of associated and more ap-parent injuries, such as pelvic fracture (40%–55%),splenic injury (60%), and renal injury (46), which mayreceive more clinical attention. Failure to identify dia-phragmatic injury may lead to intrathoracic visceralherniation and subsequent strangulation, with mortal-ity and morbidity rates as high as 50% (47).

The mechanisms of injury include (a) lateral im-pact, which distorts the chest wall and tears the dia-phragm, and (b) a direct blow to the abdomen, lead-ing to increased intraabdominal pressure and subse-quent diaphragmatic rupture (46). In contrast topenetrating injuries, injury sustained from bluntchest trauma usually produces a long tear, generallylonger than 10 cm, and occurs in the posterolateralaspect of the hemidiaphragm between its intercostalattachment and the lumbar spine (44).

Chest radiographic findings specific for diaphrag-matic rupture include intrathoracic herniation of ahollow viscus and depiction of the nasogastric tubeabove the left hemidiaphragm (44). Other radio-graphic findings that are suggestive of but not specificfor diaphragmatic injury include apparent elevation ofthe hemidiaphragm, distortion or obliteration of thediaphragmatic outline, and a contralateral shift of themediastinum (48). Even though chest radiographs arerecommended in all patients with major trauma, chestradiographs are insensitive for identifying diaphrag-matic rupture (sensitivity of 46% for the left and 17%for the right) (48). Delayed rupture of the diaphragmhas been reported in intubated patients as positivepressure ventilation is withdrawn (49).

With the advent of helical and now multi–detectorrow CT technology, the diagnostic accuracy of CT fordiaphragmatic injury has improved (50). CT has a re-ported sensitivity and specificity of 61%–71% and87%–100%, respectively, for acute traumatic dia-phragmatic rupture (51,52). CT findings suggestive ofhemidiaphragm rupture include discontinuity of thehemidiaphragm (73% sensitivity and 90% specificity)(52,53), intrathoracic herniation of abdominal con-tents (55% sensitivity and 100% specificity) (52), thecollar sign (a waistlike constriction of the herniatinghollow viscus at the site of diaphragmatic tear, with63% sensitivity and 100% specificity) (51), and thedependent viscera sign (herniated viscera layering de-pendently in the hemithorax against the posteriorribs). In one series, investigators reported a positivedependent viscera sign with 100% of left and 83% ofright hemidiaphragmatic injuries (45).

Even though CT has a higher sensitivity and speci-ficity than chest radiography for diaphragmatic injury,several pitfalls must be avoided so as not to misdiag-nose diaphragmatic injury. A Bochdalek hernia, a con-genital posterolateral diaphragmatic defect occurringpredominantly on the left and occurring in approxi-mately 6% of asymptomatic adults, can mimic dia-phragmatic rupture (54). Diaphragmatic eventrationcan also mimic diaphragmatic tear; however, coronaland sagittal reformations are helpful in demonstratingthe isolated focal elevation of the hemidiaphragmwithout discontinuity (44). When diaphragmatic rup-ture manifests with pleural effusion, the underlyingdefect may be obscured, particularly with small tearswithout associated herniation of intraabdominal con-tents (44).

Magnetic resonance (MR) imaging is not ideal forthe initial evaluation of diaphragmatic injury andshould be reserved for patients with equivocal find-ings at CT or delayed signs and symptoms of a dia-phragmatic tear (44).

CHEST WALL INJURY

Rib Fractures

Rib fractures are the most common injury followingblunt chest trauma (13,55). The most common sitesof rib fractures are ribs 4–9 laterally, where there isless overlying musculature (3,25). However, fractureof the first and/or second rib is a hallmark of high-en-ergy trauma because these ribs are short, thick, andrelatively well protected by the thoracic muscles (3).Injuries associated with first and second rib fractures

Figure 8. Chest radiograph of a 20-year-old man involved inmotor vehicle accident shows nasogastric tube (arrow) cours-ing into left hemithorax above the expected position of dia-phragm, indicating diaphragmatic rupture and herniation ofstomach into chest.

Imag

ing

of Blu

nt C

hest Trau

ma

77

include pulmonary and cardiac contusion, neck in-juries, and severe abdominal injuries (3,5). Isolatedfirst rib fractures are also associated with whiplashinjuries (56). Lacerations of the liver, spleen, andkidney are associated with fractures of ribs 9–12 (3).

Although rib fracture is a common injury, not allrib fractures are identified on the initial chest radio-graph, particularly when they are not displaced (57).CT has proved to be useful in the setting of rib frac-tures, not only because it can show nondisplacedfractures, but also because it can help to identify in-juries associated with rib fractures, such as pulmo-nary laceration or abdominal visceral injuries.

Flail chest deformity is a serious manifestation ofrib fracture and is defined as five or more adjacentrib fractures or more than three segmental rib frac-tures (9,24,55,58). Flail chest deformity can lead torespiratory failure from the direct effect of lung andpleural injury, as well as impaired ventilationcaused by dysfunction of normal chest wall mechan-ics. Although the normal hemithorax expands dur-ing inhalation, the flail hemithorax will paradoxi-cally retract, leading to hypoventilation of the af-fected lung, with rebreathing of stagnant air(9,24,55,58).

Sternal Fracture

Sternal fracture (Fig 9) occurs in about 8% of thepatients admitted for blunt chest injury (59), and themajority of such fractures occur in elderly patients.Motor vehicle accidents are the cause of about 80% ofsternal fractures (20). Sternal fracture is generally amarker for high-energy trauma and is associated withinjuries to mediastinal structures, such as the heart,great vessels, and tracheobronchial tree.

Sternal fractures cannot be seen on frontal chestradiographs. A lateral view may help to identify asternal fracture, but CT is the examination of choice,especially because it can show associated mediasti-nal injuries (13).

Sternoclavicular Dislocation

The mechanism of injury is generally a lateral com-pression force to the shoulder, with the force transmit-ted through the clavicle, or a direct blow to the medialaspect of the clavicle. The epiphysis usually remains at-tached to the sternum, but the medial aspect of the clav-icular shaft may be displaced anteriorly or posteriorly.

Anterior dislocation of the medial head of theclavicle is more common than posterior dislocation.However, posterior sternoclavicular dislocation isfar more serious because of the risk of great vesselinjury (60).

Chest radiographs provide somewhat limited de-piction of the sternoclavicular joints because of over-lying structures. Again, CT is superior to radiographyin evaluating the presence and complications of ster-noclavicular dislocation. However, the key to recog-nizing this injury is clinical evaluation, with imagingused to confirm the presence of the injury (14).

Scapular Fracture

The scapula is a well-protected structure, and,therefore, a scapular fracture is a marker of high-en-ergy trauma. On the chest radiograph, scapular frac-ture is overlooked in as many as 43% of the patients.Moreover, 72% of these unobserved fractures are vis-ible in retrospect on the initial radiograph (61). CT ismore sensitive than radiography for depicting thefracture site and its associated injuries, which includerib fractures, pneumothorax, hemothorax, and pul-monary contusion (13).

Thoracic Spine Fracture

Fracture of the thoracic spine accounts for approxi-mately 25%–30% of all spine fractures (62). It usu-ally occurs with motor vehicle accidents or with afall from great height. The mechanism of injury in-cludes hyperflexion and/or axial loading (63). Tho-racic spine fractures or dislocations have the highestincidence of associated neurologic deficits, com-pared with fractures elsewhere in the spine (11).

Figure 9. The posteroanterior chest radiograph (not shown) ofa 20-year-old man with chest pain following motor vehicle acci-dent had appeared normal. This lateral chest radiograph ob-tained on same day shows segmental fracture of lower segmentof sternum (arrow).

Tho

on

gsu

wan

et a

l

78

Chest radiography is not an adequate study to com-pletely evaluate the thoracic spine. Dedicated frontaland lateral radiographs centered on and collimated tothe thoracic spine are necessary to provide the mini-mally acceptable radiologic evaluation. Radiographicsigns of thoracic spine fracture include cortical disrup-tion, vertebral body height loss or deformity, abnor-mal vertebral alignment, focal mediastinal contourabnormality, and focal lateral displacement of theparavertebral stripe from paraspinal hematoma (8,63).CT is the imaging modality of choice for evaluatingspinal fracture because of its high sensitivity and theability to reformat images in multiple planes, particu-larly with multi–detector row CT (64).

MR imaging is a useful adjunct imaging modality toevaluate spinal soft tissues, including the intervertebraldisks, spinal ligaments, paravertebral soft tissues, spinalcord, and nerve roots. However, MR imaging does notdemonstrate actual fractures as well as conventional ra-diography or CT does (65–67), and because of the im-aging time and the difficulty with life-support equip-ment, MR is not generally used in the primary imagingevaluation.

In conclusion, chest imaging plays an importantrole in the diagnosis of blunt chest trauma, becausethe history and the findings from physical examina-tion are often unreliable. However, each imaging mo-dality has its limitations. Selecting the appropriatestudy for the suspected diagnosis is very important. Fi-nally, the radiologist should be aware of the limitationsof the supine chest radiograph, to minimize overdiag-nosis or underdiagnosis.

References1. Keough V, Pudelek B. Blunt chest trauma: review of se-

lected pulmonary injuries focusing on pulmonary contusion.AACN Clin Issues 2001; 12:270–281.

2. Bowling WM, Wilson RF, Kelen GD, Buchman TG. Thoracictrauma. In: Tintnialli JE, Kelen GD, Stapczynski JS, eds.Emergency medicine. New York, NY: McGraw-Hill, 2000;251.

3. Cogbill TH, Landercasper J. Injury to the chest wall. In:Mattox KL, Felciano DV, Moore EE, eds. Trauma. NewYork, NY: McGraw-Hill, 2000; 483–504.

4. National Safety Council. Report on injuries in America 2000.Available at: www.nsc.org/library/rept2000.htm. AccessedJune 15, 2004.

5. Miller FB, Richardson JD, Thomas HA, Cryer HM, WillingSJ. Role of CT in diagnosis of major arterial injury afterblunt thoracic trauma. Surgery 1989; 106:596–602.

6. Mirvis SE, Tobin KD, Kostrubiak I, Belzberg H. Thoracic CTin detecting occult disease in critically ill patients. AJR Am JRoentgenol 1987; 148:685–689.

7. Brooks SW, Young JC, Cmolik B, et al. The use of trans-esophageal echocardiography in the evaluation of chesttrauma. J Trauma 1992; 32:761–765.

8. Groskin SA. Selected topics in chest trauma. Radiology1992; 183:605–617.

9. Greene R. Blunt thoracic trauma. In: Greene R, Muhm JR,eds. Syllabus: a categorical course in diagnostic radiol-ogy—chest radiology. Oak Brook, Ill: Radiological Societyof North America, 1992; 297–309.

10. Williams JR, Stembridge VA. Pulmonary contusion second-ary to nonpenetrating chest trauma. Am J Roentgenol Ra-dium Ther Nucl Med 1964; 91:284–290.

11. Van Hise ML, Primack SL, Israel RS, Müller NL. CT in bluntchest trauma: indications and limitations. RadioGraphics1998; 18:1071–1084.

12. Wagner RB, Crawford WO Jr, Schimpf PP. Classification ofparenchymal injuries of the lung. Radiology 1988; 167:77–82.

13. Kerns SR, Gay SB. CT of blunt chest trauma. AJR Am JRoentgenol 1990; 154:55–60.

14. Dee PM. The radiology of chest trauma. Radiol Clin NorthAm 1992; 30:291–306.

15. Besson A, Saegesser FA. Lung injury from blunt chesttrauma. In: Color atlas of chest trauma and associated inju-ries. Weert, the Netherlands: Wolfe, 1982; 156–166.

16. Greenberg MD, Rosen CL. Evaluation of the patient withblunt chest trauma: an evidence based approach. EmergMed Clin North Am 1999; 17:41–62.

17. Dee P. Chest trauma. In: Armstrong P, Wilson AG, Dee P,Hansell DM, eds. Imaging of diseases of the chest. 3rd ed.London, England: Mosby, 2000; 949–981.

18. Ashbaugh DG, Peters GN, Halgrimson CG, Owens JC,Waddell WR. Chest trauma: analysis of 685 patients. ArchSurg 1967; 95:546–555.

19. Conn JH, Hardy JD, Fain WR, Netterville RE. Thoracictrauma: analysis of 1022 cases. J Trauma 1963; 3:22–40.

20. Chan O, Hiorns M. Chest trauma. Eur J Radiol 1996; 23:23–34.

21. Tocino IM, Miller MH, Frederick PR, Bahr AL, Thomas F.CT detection of occult pneumothorax in head trauma. AJRAm J Roentgenol 1984; 143:987–990.

22. Wall SD, Federle MP, Jeffrey RB, Brett CM. CT diagnosisof unsuspected pneumothorax after blunt abdominaltrauma. AJR Am J Roentgenol 1983; 141:919–921.

23. Ziter FM Jr, Westcott JL. Supine subpulmonary pneumotho-rax. AJR Am J Roentgenol 1981; 137:699–701.

24. Kazerooni EA, Gross BH. Thoracic trauma. In: KazerooniEA, Gross BH, eds. Core curriculum: cardiopulmonary im-aging. Philadelphia, Pa: Lippincott Williams & Wilkins, 2004;295–322.

25. Richardson JD, Spain DA. Injury to the lung and pleura. In:Mattox KL, Feliciano DV, Moore EE, eds. Trauma. NewYork, NY: McGraw-Hill, 2000; 523–543.

26. Bertelsen S, Howitz P. Injuries of the trachea and bronchi.Thorax 1972; 27:188–194.

27. Wiot JF. Tracheobronchial trauma. Semin Roentgenol 1983;18:15–22.

28. Kirsh MM, Orringer MB, Behrendt DM, Sloan H. Manage-ment of tracheobronchial disruption secondary to non-penetrating trauma. Ann Thorac Surg 1976; 22:93–101.

29. Burke JF. Early diagnosis of traumatic rupture of the bron-chus. JAMA 1962; 181:682–686.

30. Lee RB. Traumatic injury of the cervicothoracic trachea andmajor bronchi. Chest Surg Clin N Am 1997; 7:285–304.

31. Ketai L, Brandt MM, Schermer C. Nonaortic mediastinal in-juries from blunt chest trauma. J Thorac Imaging 2000; 15:120–127.

32. Oh KS, Fleischner FG, Wyman SM. Characteristic pulmo-nary finding in traumatic complete transection of a main-stem bronchus. Radiology 1969; 92:371–372.

33. Kumpe DA, Oh KS, Wyman SM. A characteristic pulmonaryfinding in unilateral complete bronchial transection. Am JRoentgenol Radium Ther Nucl Med 1970; 110:704–706.

34. Petterson C, Deslauriers J, McClish A. A classic image ofcomplete right main bronchus avulsion. Chest 1989; 96:1415–1417.

35. Unger JM, Schuchmann GG, Grossman JE, Pellett JR.Tears of the trachea and main bronchi caused by blunttrauma: radiologic findings. AJR Am J Roentgenol 1989;153:1175–1180.

Imag

ing

of Blu

nt C

hest Trau

ma

79

36. Lupetin AR. Computed tomographic evaluation of laryngotra-cheal trauma. Curr Probl Diagn Radiol 1997; 26:185–206.

37. Palder SB, Shandling B, Manson D. Rupture of the thoracictrachea following blunt trauma: diagnosis by CAT scan. JPediatr Surg 1991; 26:1320–1322.

38. d’Odemont JP, Pringot J, Goncette L, Goenen M, Roden-stein DO. Spontaneous favorable outcome of tracheal lacera-tion. Chest 1991; 99:1290–1292.

39. Baumgartner FJ, Ayres B, Theuer C. Danger of false intu-bation after traumatic tracheal transection. Ann Thorac Surg1997; 63:227–228.

40. Chen JD, Shanmuganathan K, Mirvis SE, Killeen KL,Dutton RP. Using CT to diagnose tracheal rupture. AJR AmJ Roentgenol 2001; 176:1273–1280.

41. Eijgelaar A, Homan van der Heide JN. A reliable earlysymptom of bronchial or tracheal rupture. Thorax 1970;25:120–125.

42. Spencer JA, Rogers CE, Westaby S. Clinico-radiologicalcorrelates in rupture of the major airways. Clin Radiol 1991;43:371–376.

43. Lotz PR, Martel W, Rohwedder JJ, Green RA. Significanceof pneumomediastinum in blunt trauma to the thorax. AJRAm J Roentgenol 1979; 132:817–819.

44. Iochum S, Ludig T, Walter F, Sebbag H, Grosdidier G, BlumAG. Imaging of diaphragmatic injury: a diagnostic chal-lenge? RadioGraphics 2002; 22(spec no):S103–S116.

45. Bergin D, Ennis R, Keogh C, Fenlon HM, Murray JG. The"dependent viscera" sign in CT diagnosis of blunt traumaticdiaphragmatic rupture. AJR Am J Roentgenol 2001;177:1137–1140.

46. Shanmuganathan K, Killeen K, Mirvis SE, White CS. Imag-ing of diaphragmatic injuries. J Thorac Imaging 2000; 15:104–111.

47. Drews JA, Mercer EC, Benfield JR. Acute diaphragmatic in-juries. Ann Thorac Surg 1973; 16:67–78.

48. Gelman R, Mirvis SE, Gens D. Diaphragmatic rupture dueto blunt trauma: sensitivity of plain chest radiographs. AJRAm J Roentgenol 1991; 156:51–57.

49. Carter YM, Karmy-Jones RC, Stern EJ. Delayed recognitionof diaphragmatic rupture in a patient receiving mechanicalventilation. AJR Am J Roentgenol 2001; 176:428.

50. Blum A, Walter F, Ludig T, Zhu X, Roland J. Multislice CT:principles and new CT-scan applications. J Radiol 2000; 81:1597–1614.

51. Killeen KL, Mirvis SE, Shanmuganathan K. Helical CT ofdiaphragmatic rupture caused by blunt trauma. AJR Am JRoentgenol 1999; 173:1611–1616.

52. Murray JG, Caoili E, Gruden JF, Evans SJ, Halvorsen RAJr, Mackersie RC. Acute rupture of the diaphragm due toblunt trauma: diagnostic sensitivity and specificity of CT.AJR Am J Roentgenol 1996; 166:1035–1039.

53. Worthy SA, Kang EY, Hartman TE, Kwong JS, Mayo JR,Muller NL. Diaphragmatic rupture: CT findings in 11 pa-tients. Radiology 1995; 194:885–888.

54. Caskey CI, Zerhouni EA, Fishman EK, Rahmouni AD. Agingof the diaphragm: a CT study. Radiology 1989; 171:385–389.

55. Kuhlman JE, Pozniak MA, Collins J, Knisely BL. Radio-graphic and CT findings of blunt chest trauma: aortic inju-ries and looking beyond them. RadioGraphics 1998;18:1085–1108.

56. Qureshi T, Mander BJ, Wishart GC. Isolated bilateral firstrib fractures: an unusual sequel of whiplash injury. Injury1998; 29:397–398.

57. Tocino I, Miller MH. Computed tomography in blunt chesttrauma. J Thorac Imaging 1987; 2:45–59.

58. Gurney JW. ABCs of blunt chest trauma. In: Thoracic imag-ing. Reston, Va: Society of Thoracic Radiology, 1996; 349–352.

59. Fisher RG, Ward RE, Ben-Menachem Y, Mattox KL, FlynnTC. Arteriography and the fractured first rib: too much fortoo little? AJR Am J Roentgenol 1982; 138:1059–1062.

60. Gerlock AJ Jr, Muhletaler CA, Coulam CM, Hayes PT.Traumatic aortic aneurysm: validity of esophageal tube dis-placement sign. AJR Am J Roentgenol 1980; 135:713–718.

61. Harris RD, Harris JH Jr. The prevalence and significance ofmissed scapular fractures in blunt chest trauma. AJR Am JRoentgenol 1988; 151:747–750.

62. Pal JM, Mulder DS, Brown RA, Fleiszer DM. Assessingmultiple trauma: is the cervical spine enough? J Trauma1988; 28:1282–1284.

63. Daffner RH. Imaging of vertebral trauma. Rockville, Md: As-pen, 1988.

64. el-Khoury GY, Whitten CG. Trauma to the upper thoracicspine: anatomy, biomechanics, and unique imaging fea-tures. AJR Am J Roentgenol 1993; 160:95–102.

65. Goldberg AL, Rothfus WE, Deeb ZL, et al. The impact ofmagnetic resonance on the diagnostic evaluation of acutecervicothoracic spinal trauma. Skeletal Radiol 1988; 17:89–95.

66. Kalfas I, Wilberger J, Goldberg A, Prostko ER. Magneticresonance imaging in acute spinal cord trauma. Neurosur-gery 1988; 23:295–299.

67. Mathis JM, Wilson JT, Barnard JW, Zelenik ME. MR imag-ing of spinal cord avulsion. AJNR Am J Neuroradiol 1988;9:1232–1233.

NO

TES

81

Imaging Diagnosis ofThoracic Aorta and

Great Vessel Injuries1

The rapid diagnosis of traumatic injury to the aorta and its major branches follow-ing major blunt trauma is both extremely urgent and potentially challenging. Thecombination of the relative rarity of the injury (<0.5% of major trauma admissions)and the high lethality when not treated (40% mortality at 24 hours after injury) cre-ates this demanding clinical situation (1). The newer imaging modalities, includingmulti–detector row helical computed tomography (CT), intravascular and transesoph-ageal ultrasonography (US), and, occasionally, magnetic resonance imaging, havebeen key factors in establishing the diagnosis with higher accuracy and in less timethan had been true prior to application of these techniques.

Still, in most centers, the chest radiograph continues to serve as the initial diag-nostic study for detecting injury within the mediastinum. This study remains sensi-tive for the detection of mediastinal hemorrhage and potential aortic injury but isimpaired by low specificity (2). Multi–detector row CT has rapidly become the sec-ondary study performed in patients with abnormalities of the mediastinal contour(3–6) and in many institutions has replaced the previous reference standard of tho-racic angiography. As described in this chapter, the high sensitivity of multi–detectorrow CT has increased the recognized spectrum of injuries to the aorta and great vesselsand has opened the door for selective nonsurgical management of some injuries.

HOW TO USE THE CHEST RADIOGRAPH

Most patients who have undergone acute blunt trauma whose condition is hemody-namically stable will undergo chest radiography among several routine studies per-formed at admission. The chest radiograph can be used to detect a number of immedi-ately life-threatening traumatic pathologic conditions, including gross hemothorax,tension pneumothorax, tension pneumopericardium, and ruptured diaphragm withmassive herniation of abdominal viscera. Careful inspection of the mediastinum is re-quired to detect contour abnormalities that suggest hemorrhage. During the past de-cades, many radiologic findings have been described to indicate a high likelihood ofmediastinal hematoma (7–10). A mediastinal diameter at the level of the aortic archof greater than 8 cm (widened mediastinum) and a mediastinal–to–chest-width ratioof more than 0.25% at the same level have been popular, though unreliable, indica-tors for further investigation for potential aortic injury (11).

Stuart E. Mirvis, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 81–89.

1From the Department of Radiology and the Maryland Shock-Trauma Center, University of Maryland School of Medicine,22 S Greene St, Baltimore, MD 21201-1544 (e-mail: smirvis@umm.edu).

Mir

vis

82

The disturbance of the normal shadows of the medi-astinal contour, such as an obscured aortic arch or de-scending aorta or an abnormal contour of the arch, theloss of the aortopulmonary window, widening of theleft paraspinal stripe or its extension to the left extra-pleural apex, and right paratracheal soft-tissue densityaccurately indicate potential mediastinal hematoma(12,13) (Figs 1, 2). Unfortunately, this approach re-quires a well-established knowledge of radiographicanatomic structures, experience in interpreting a largenumber of chest radiographs, and specific experiencewith radiographs of patients with mediastinal hemor-rhage. The results of large series have verified the lim-ited utility of simple quantitative measures of the medi-astinal diameter to guide management (11,14).

The presence of a mediastinal contour abnormalityis still a nonspecific finding for major thoracic arterialinjury. A number of reasons account for this limita-tion, among them the following: (a) limited qualityof portable radiographs, (b) suboptimal or lack of pa-tient cooperation, (c) magnification and distortion ofmediastinal silhouettes related to the supine view, and(d) nontraumatic causes of mediastinal contour ab-normality, such as lymphadenopathy, mediastinal li-pomatosis, vascular ectasia, and aberrant vascularbranching (15). In many cases, the mediastinal con-tour is effaced by both traumatic and nontraumaticentities. In trauma, common sources of obscuration ofthe mediastinal margins include atelectasis, medialpleural effusions, and lung contusions and hemato-mas. Another serious limitation of chest radiographyin the diagnosis of traumatic aortic injury is the fact

that most patients (approximately 80%) with medias-tinal blood do not actually have a major thoracic arte-rial injury (12). Thus, even if all patients with true me-diastinal hemorrhage could be selected with radiogra-phy, the true-positive rate for vascular injury wouldonly approach 20% (12).

The use of the erect or true erect (leaning forward15°) view can improve the specificity of the chest ra-diograph and can often show a normal mediastinalcontour when the supine view appears abnormal(16). A chest radiograph with a normal mediastinalcontour for age has at least a 98% negative predictivevalue for traumatic aortic injury (2,12). Again, the ac-curate interpretation of the chest radiograph and thedecision to accept it as showing a normal mediasti-num or to go on to further imaging is vitally impor-tant and requires both experience and confidence.

EMERGENCE OF THORACIC CT IN DIAGNOSISOF MAJOR VASCULAR INJURY

Beginning in 1983, investigators in numerous studieshave described the use of CT to diagnose aortic injury.Of course, in the earliest studies, the investigatorsstudied relatively small series and used the relativelyslow, low-resolution conventional CT systems of themid-1980s (17–19). Naturally, these early attempts at“CT angiography” with an 8–10-mm section thicknesswere limited but still clearly showed that aortic inju-ries could be diagnosed with CT. The issue of sensitiv-ity was still open to question.

During the next 15 years, debate raged among sur-geons and cross-sectional imagers about the true accu-

Figure 1. Normalmediastinal con-tours. Anteropos-terior view of chestobtained at time ofadmission of a pa-tient with blunttrauma shows awell-defined aorticarch and descend-ing aorta, no softtissue in rightparatracheal re-gion, no evidenceof widening ofparaspinal stripes,and a midline tra-chea. These find-ings together havea greater than 98%negative predictivevalue for majormediastinal arterialinjury.

Figure 2. Medi-astinal hemor-rhage. Anteropos-terior chest radio-graph of patientwith blunt traumashows abnormalcontour of aorticarch and trachealdeviation to right(arrow). Findingsindicate mediasti-nal hemorrhage inthis setting andwere associatedwith aortic injury inthis case.

Tho

racic Ao

rta and

Great V

essel Inju

ries

83

racy of CT and whether high accuracy could be achievedroutinely or only in specialized trauma centers (20–23).The time-honored reference standard of thoracic angiog-raphy remained firmly in place in many institutions. Anumber of issues crystallized as this debate continued. Itwas obvious that angiography was invasive, costly, anddifficult to perform quickly. In most polytrauma cases,CT scans were indicated for assessment of other bodyregions. Finally, it became clear with careful review thatthoracic angiography for aortic injury was often chal-lenging to interpret and occasionally incorrect diagnos-tically (24,25). With the advent of multi-detector rowCT and the potential to create angiogram-like images ofthe aorta and its major branches, the pendulum formost trauma centers swung toward multi–detector rowCT as the test of choice in stable patients with blunttrauma who do not have a clearly normal mediastinalcontour at chest radiography.

CT IN AORTIC AND MAJOR BRANCH VESSELINJURY

The diagnostic accuracy of CT for thoracic aortic andmajor branch vessel injury steadily improves as onemoves from single-section helical to 4–detector rowCT and, subsequently, 16–detector row CT. At eachstep, image quality is improved both with the use ofthinner axial sections and with decreased motion arti-fact. The quality of multiplanar reformations in anyorthogonal axis or in the curved axis of the aorta andits first-order branches improves similarly with the useof thin (1–2-mm) and overlapping axial sections.

Figure 3. Traumatic aortic injury withperidiaphragmatic hemorrhage. (a) Supinechest radiograph obtained at admissiondemonstrates wide mediastinal shadow withno depiction of aortic arch or descendingaorta, a right paratracheal density, and tra-cheal deviation to right, indicating mediasti-nal blood. (b) Axial multi–detector row CTimage through aortic hiatus shows hemor-rhage around aorta at this level. (c) Axial CTimage through proximal descending aortashows diffuse mediastinal hemorrhage, dis-placement of carina to right, and aorticpseudoaneurysm arising from medial aspectof aorta.

In my practice, thoracic CT is performed with a 16–detector row CT scanner with a 16 × 1.5-mm detectorarray, a 0.75-second scan time, and pitch varying from1 to 1.2. Axial images are usually viewed with fusionof three 1.5-mm images, which are subsequentlysaved to the picture archiving and communicationssystem. If needed, the axial 1.5-mm images are avail-able for 3 days online. All reformations are acquiredby using the original 1.5-mm axial images. All tho-racic studies are performed with intravenous contrastmaterial enhancement with automated bolus trigger-ing. The chest images are routinely viewed in soft-tissue, lung, and bone windows. For general cases ofblunt trauma, reformations are not obtained unlessthe study is positive and further elucidation of theinjury is needed.

Almost all traumatic aortic injuries are associatedwith hemorrhage around the aorta or its proximalbranches. The quantity of blood can range from mini-mal and focal to abundant and diffuse. Often, periaor-tic hemorrhage will track along the vessel inferiorlythrough the aortic hiatus and along the abdominalaorta (Fig 3). This is one sign seen at abdominal CTthat should initiate a mandatory evaluation of thethoracic aorta.

A number of direct signs of aortic and proximalbranch vessel injury can be observed at contrast mate-rial–enhanced multi–detector row CT (3–5,25–27). Inthe most typical case, a pseudoaneurysm or containedpartial aortic wall tear is seen projecting anteriorly atthe level of the left main stem bronchus and left pul-monary artery (Figs 3–5). This region is believed to beprone to injury through shearing effects and stretch-ing, compression, and torsion (28,29). In addition, therelative disorganization of the elastic lamina at the levelof the remnant ductus arteriosus may create a localweakness in the wall at this level. The pseudoaneurysmis usually delineated by intimal flaps on either side andmay be larger in size than the true aortic lumen (Fig 6).The pseudoaneurysm is usually relatively small in lon-gitudinal extent and diameter (1–3 cm), forming anacute angle with the aortic wall and with a smooth ex-terior surface (Figs 3, 5).

Mir

vis

84

The second most common site of injury is the as-cending aorta just above the valve ring (Fig 7). Injuryat this site is seldom seen clinically because of itshigh early mortality. Other sites of potential injuryinclude the aortic arch, arch–branch vessel origins,and the descending aorta (Figs 5, 6). In my experi-ence, injuries to the arch itself are relatively morecommon in elderly individuals with ectatic aortasthan in a younger population.

When the aortic pseudoaneurysm is large, it mayindent the lumen of the aorta and produce a pseudo-coarctation, leading to decreased perfusion belowthe level of injury. This finding is manifest as a sud-den decrease in the diameter of the aorta along a seg-ment without major branches. The descending andabdominal aorta will appear smaller in caliber thanwould be expected (Figs 8, 9). Clinically, there willbe a decrease in blood pressure and pulse strength inthe lower extremity. A small abdominal aorta is an-other CT sign that should prompt assessment of thethorax in a patient with blunt trauma. Other signs ofdirect aortic injury at CT include an irregularlyshaped lumen, one or more linear filling defects dueto intimal flaps, intraluminal blood clot, suddennarrowing of the luminal caliber as noted previously,and, rarely, extravasation of intravenous contrast ma-terial into the periaortic hematoma (Figs 3–6, 9, 10).

In most cases, these injuries are apparent on high-quality multi–detector row systems. If there are no di-rect findings on the axial images but there is bloodaround the aorta or great vessels, then thin-sectionmultiplanar reformations along the aortic and greatvessel axes should be obtained because these can en-hance appreciation of injuries that may be subtle find-ings on axial views alone (Fig 11). When there are di-rect findings of aortic and branch vessel injury, furtherassessment with multiplanar reformations along themajor vascular axes is also valuable to demonstrate theinjury in relation to other vessels. On occasion, I have

found endoluminal (“angioscopic”) views helpful inverifying subtle injuries (Fig 12).

The term traumatic aortic dissection is not appropri-ate for most aortic injuries. These injuries may con-tain a short interval when blood under pressure dis-sects a short distance into an otherwise normal aorticmedia, but this is not a principal feature of the in-jury. In rare cases, I have seen long medial or subad-ventitial dissections within a normal aorta, resultingfrom blunt trauma, that extend from the thoracicaortic tear into the abdominal aorta (Fig 13), butthese are exceptional injuries.

To be of true value, the multi–detector row CT studyof the aorta should not only demonstrate the presenceof an injury, but also characterize its severity, size, andextent. A small intimal irregularity may be managed ef-

Figure 4. Aorticpseudoaneurysm.Axial multi–detec-tor row CT imageof 42-year-oldwoman who wasinjured in a high-speed motor ve-hicular crashshows pseudo-aneurysm arisingfrom anterior as-pect of proximaldescending aorta(arrow) at left ofleft main pulmo-nary artery. Note moderate surrounding mediastinal hemor-rhage. Pseudoaneurysm is larger than aortic lumen and com-presses it.

Figure 5. Aorticpseudoaneurysm.Axial CT image ofanother patientwith blunt chesttrauma showstypical pseudo-aneurysm (arrow)arising from ante-rior proximal de-scending aorta atlevel of left mainstem bronchus.

Figure 6. Largeaortic pseudo-aneurysm and inti-mal flap. Axial CTimage of a 66-year-old manstruck by a carshows hugepseudoaneurysm(P) arising fromanterior proximaldescending aorta.Flap of disruptedintima (arrow) di-vides pseudo-aneurysm from na-tive aortic lumen. Note diffuse mediastinal blood.

Figure 8. Pseudo-coarctation of in-jured aorta. AxialCT image throughupper abdomen ofsame patient as inFigure 4 showssmall-caliber aortasecondary to com-pression of aorticlumen by thoracicaortic pseudo-aneurysm.

Tho

racic Ao

rta and

Great V

essel Inju

ries

85

fectively with blood pressure control and observation,whereas large pseudoaneurysms will typically be man-aged surgically or, in some centers, with endovascularstent placement (30,31). It is vital to show the relation-ship of the injury to adjacent branches, the branchingpattern that will be encountered at surgery, and the ex-

Figure 9. Pseudocoarctation of injuredaorta. (a) Axial CT image through aortic injuryin a 34-year-old man injured in motorcyclecrash shows area of soft-tissue attenuationacross the aorta, consistent with thrombus.(b) Axial CT image of upper portion of abdo-men shows small aortic luminal diametersecondary to decreased flow through the in-jured area in the proximal thoracic aorta.Note well-defined low-attenuation areas inspleen, most likely representing infarcts fromembolization from aortic thrombus.

Figure 7. Ascending aortic pseudoaneurysm. (a) Axial multi–detector row CT imageshows three lumina arising from the base of the heart. Middle "lumen" is pseudo-aneurysm (P) arising from proximal ascending aorta. (b) Coronal multiplanar reforma-tion shows relation of pseudoaneurysm (arrow) to proximal aorta and pulmonary artery.Injury was successfully repaired. (c) Angiographic image shows ascending aortic pseudoaneurysm. (Parts a and b: Reprinted, withpermission, from reference 38.).

istence of more than one site of injury. Although a typi-cal repair of the proximal descending aorta will be per-formed through a left posterior thoracotomy, an archor proximal branch vessel injury requires a median ster-notomy for adequate exposure, but would be subopti-mal for the typical injury site in the proximal descend-ing thoracic aorta. Multiplanar reformation, surfacecontour, and volumetric images are usually best to dis-play the relationship of an aortic or branch vessel in-jury to the surrounding vascular and nonvascular struc-tures. The precise distance from the edges of the injuryto adjacent arterial branches is helpful in planningendovascular stent placement, when appropriate.

SHOULD AORTOGRAPHY BE PERFORMED?

The use of aortography has diminished with the ascentof multi–detector row CT for trauma imaging. Still, aor-tography can play important selective roles. If a good-quality multi–detector row CT study shows normal vas-culature without hemorrhage around the major tho-racic arteries, no further studies are required. In cases inwhich perivascular blood and normal vessels are seen

Figure 10. Aorticinjury. Axial CT im-age through aorticinjury showsslightly irregularaortic lumen, peri-aortic mediastinalhemorrhage, andthrombus in lumen.Note displacementof nasogastric tubeto right.

Mir

vis

86

at multi–detector row CT, the decision of how to pro-ceed is based on experience. Individuals with experi-ence in chest CT interpretation for trauma may confi-dently consider the study normal for the vessels andnot proceed to further investigation. Those with lesserexperience may opt to perform other studies, such asaortography or intravascular or transesophageal US(32). The decision of which study to perform shouldbe made on the basis of the experience of the exam-iner and the availability of the study. It should rarelybe necessary to perform an “exploratory thoracoto-my,” given current diagnostic capabilities. In patientswith studies that are unequivocally positive for one ormore injuries, it should not be necessary to performaortography, which may introduce potentially longdelays in initiating surgical care.

Blood pressure control should be an immediate stepin management, once a possible aortic injury is seen,but this method is not a foolproof way to avoid a sud-den complete tear, free hemorrhage, and rapid exsan-guination. If there is any question of active bleedingfrom the aorta into the mediastinal hematoma, imme-diate thoracic surgery should be performed becausethis situation can rapidly become fatal (33). It seemsprudent that the urgency for a definitive diagnosis ofmajor intrathoracic branch vessel injury should be thesame as that for injury of the thoracic aorta, althoughthe natural history of these injuries is less well known.

PITFALLS OF MULTI–DETECTOR ROW CT INAORTIC INJURY

As in all diagnostic studies, the combination of in-creased use and familiarity with the method uncoversnew difficulties. Technical problems in performing

contrast-enhanced thoracic CT will limit the utility ofthe study, but even with a complete lack of intrave-nous contrast material, this study should allow the di-agnosis or exclusion of mediastinal hemorrhage andthus lead to the correct next step.

A number of situations can confound interpreta-tion of even well-performed multi–detector row CTstudies. Among these are (a) subtle injuries, (b) abackground of severe atherosclerotic disease with ul-ceration, (c) variants in vascular branching (Fig 14),and (d) diverticular bronchial artery origin, as well as(e) the ductus diverticulum (24,25). Similarly, a diver-ticular origin of the bronchial artery can mimic an in-jured aorta. A traumatic pseudoaneurysm and penetrat-ing aortic ulceration could appear similar, although theformer is typically in the proximal descending aortaand the latter in the midportion of the descendingaorta. Penetrating aortic ulceration is uniformly asso-ciated with severe aortic arteriosclerosis and calcifica-tion and is essentially a disease of the 7th, 8th, and9th decades of life (34). Thoracic angiographic studiescan also be difficult to interpret with some of these en-tities and may entirely miss an injury that is not shownin profile. Again, multi–detector row CT is complemen-tary with angiography and US, and these studies can behelpful if the multi–detector row CT is equivocal.

HOW WILL IMAGING AFFECT THE FUTURE OFTREATMENT?

Just as the diagnostic capabilities for major thoracic arte-rial injuries have grown markedly in the past decade, sohas the range of treatment options. Traditionally, mostcenters will urgently undertake surgical repair of thesearterial injuries, with partial or complete bypass, to

Figure 11. Subtle CT findings of traumatic aortic injury. (a, b) Two axial CT images across descending aorta show marked differencein caliber and aortic contour abnormality in a. Note periaortic hemorrhage. (c) Volume-rendered image shows slight enlargement ofaorta in proximal descending portion and intimal flaps (arrowheads). Findings of aortic injury are subtle in this case. (Part c: Reprinted,with permission, from reference 38.)

Tho

racic Ao

rta and

Great V

essel Inju

ries

87

Figure 12. Use of endoluminal view to diagnose aortic injury. (a, b) Axial CT imagesthrough proximal descending aorta demonstrate periaortic mediastinal hemorrhage andpossible wall irregularity of anterior aorta in b. (c) Endoluminal rendering shows clearintimal flap (arrow) projecting into lumen. H = oriented toward head. (d) Aortogram con-firms aortic pseudoaneurysm in typical location.

Figure 13. Dissection associated with aortic in-jury. (a) Axial CT image of young woman with blunttrauma shows ring of hemorrhage surroundingopacified lumen, with small amount of contrast ma-terial (arrow) projecting into hematoma. (b) Axial CTimage at more caudal level shows an apparentthrombosis of false lumen and narrowed true lumen.(c) Abdominal aortogram shows irregularity of pos-terior aortic wall, narrowed distal abdominal aorticlumen, and small amount of bleeding into distalfalse channel (arrow).

Mir

vis

88

9. Peters DR, Gamsu G. Displacement of the right para-spinous interface: a radiographic sign of acute traumaticrupture of the thoracic aorta. Radiology 1980; 134:599–603.

10. Marnocha KE, Maglinte DD. Plain-film criteria for excludingaortic rupture in blunt chest trauma. AJR Am J Roentgenol1985; 144:19–21.

11. Marnocha KE, Maglinte DD, Woods J, Goodman M,Peterson P. Mediastinal-width/chest-width ratio in bluntchest trauma: a reappraisal. AJR Am J Roentgenol 1984;142:275–277.

12. Mirvis SE, Bidwell JK, Buddemeyer EU. Imaging diagnosisof traumatic aortic rupture: a review and experience at amajor trauma center. Invest Radiol 1987; 22:187–196.

13. Marnocha KE, Maglinte DD, Woods J. Blunt chest traumaand suspected aortic rupture: reliability of chest radiographfindings. Ann Emerg Med 1985; 14:644–649.

14. Woodring JH, Dillon ML. Radiographic manifestations ofmediastinal hemorrhage from blunt chest trauma. AnnThorac Surg 1984; 37:171–178.

15. Schwab CW, Lawson RB, Lind JF, Garland LW. Aortic injury:comparison of supine and upright portable chest films toevaluate the widened mediastinum. Ann Emerg Med 1984;13:896–899.

Figure 14. Traumatic aortic injury with dis-section. (a) Axial CT image of patient withblunt trauma shows large pseudoaneurysmarising from anterior aspect of aortic isth-mus region, with surrounding periaortichemorrhage. (b) Axial CT image shows thataberrant right subclavian artery (arrow)crosses behind trachea at more rostrallevel. (c) Posterior volumetric view of aorticarch shows relationship between pseudo-aneurysm (Ps) and aberrant artery above it.RSC = right subclavian artery.

avoid sudden complete vessel wall rupture. More re-cently, nonsurgical management with blood pressurecontrol and serial observation has been shown to be apossible alternative (35,36). In addition, endovascularstent placement, to protect the injured area while main-taining major branch perfusion, has been used with in-creasing frequency, either as a temporizing or definitivetreatment (30,31). Although several clinical factors af-fect the decision as to what approach to use, as do thelocal facilities and expertise, this decision is also influ-enced by the nature and extent of the injury. In the fu-ture, the precise grading of major arterial injuries withmulti–detector row CT, as has been proposed previ-ously (37), should be performed in an effort to scalemanagement of a given injury to an appropriate level ofintervention to achieve successful long-term treatment.

References1. Parmley LF, Marion WC, Jahnke EJ. Nonpenetrating trau-

matic injury of the aorta. Circulation 1958; 17:1086–1092.2. Mirvis SE, Bidwell JK, Buddemeyer EU, et al. Value of

chest radiography in excluding traumatic aortic rupture. Ra-diology 1987; 163:487–493.

3. Wintermark M, Wicky S, Schnyder P. Imaging of acute trau-matic injuries of the thoracic aorta. Eur Radiol 2002; 12:431–442.

4. Melton SM, Kerby JD, McGiffin D, et al. The evolution ofchest computed tomography for the definitive diagnosis ofblunt aortic injury: a single-center experience. J Trauma2004; 56:243–250.

5. Mirvis SE, Shanmuganathan K, Buell J, Rodriguez A. Useof spiral computed tomography for the assessment of blunttrauma patients with potential aortic injury. J Trauma 1998;45:922–930.

6. Dyer DS, Moore EE, Mestek MF, et al. Can chest CT beused to exclude aortic injury? Radiology 1999; 213:195–202.

7. Simeone JF, Minagi H, Putman CE. Traumatic disruption ofthe thoracic aorta: significance of the left apical extrapleuralcap. Radiology 1975; 117:265–268.

8. Wales LR, Morishima MS, Reay D, Johansen K. Naso-gastric tube displacement in acute traumatic rupture of thethoracic aorta: a postmortem study. AJR Am J Roentgenol1982; 138:821–823.

Tho

racic Ao

rta and

Great V

essel Inju

ries

89

16. Ayella RJ, Hankins JR, Turney SZ, Cowley RA. Rupturedthoracic aorta due to blunt trauma. J Trauma 1977; 17:199–205.

17. Kubota RT, Tripp MD, Tisnado J, Cho SR. Evaluation of trau-matic rupture of descending aorta by aortography and com-puted tomography: case report with follow-up. J ComputTomogr 1985; 9:237–240.

18. Heiberg E, Wolverson MK, Sundaram M, Shields JB. CT inaortic trauma. AJR Am J Roentgenol 1983; 140:1119–1124.

19. Mirvis SE, Kostrubiak I, Whitley NO, Goldstein LD,Rodriguez A. Role of CT in excluding major arterial injuryafter blunt thoracic trauma. AJR Am J Roentgenol 1987;149:601–605.

20. Durham RM, Zuckerman D, Wolverson M, et al. Computedtomography as a screening exam in patients with suspectedblunt aortic injury. Ann Surg 1994; 220:699–704.

21. Miller FB, Richardson JD, Thomas HA, Cryer HM, WillingSJ. Role of CT in diagnosis of major arterial injury afterblunt thoracic trauma. Surgery 1989; 106:596–602.

22. Fenner MN, Fisher KS, Sergel NL, Porter DB, MetzmakerCO. Evaluation of possible traumatic thoracic aortic injuryusing aortography and CT. Am Surg 1990; 56:497–499.

23. Raptopoulos V, Sheiman RG, Phillips DA, Davidoff A, SilvaWE. Traumatic aortic tear: screening with chest CT. Radiol-ogy 1992; 182:667–673.

24. Mirvis SE, Pais SO, Shanmuganathan K. Atypical results ofthoracic aortography performed to exclude aortic rupture.Emerg Radiol 1998; 1:42–46.

25. Fisher RG, Sanchez-Torres M, Wingham CJ, et al. "Lumps"and “bumps” that mimic acute aortic and brachiocephalicvessel injury. RadioGraphics 1997; 17:825–834.

26. Cleverley JR, Barrie JR, Raymond GS, Primack SL, MayoJR. Direct findings of aortic injury on contrast-enhanced CTin surgically proven traumatic aortic injury: a multi-centre re-view. Clin Radiol 2002; 57:281–286.

27. Scaglione M, Pinto A, Pinto F, et al. Role of contrast-en-hanced helical CT in the evaluation of acute thoracic aortic

injuries after blunt chest trauma. Eur Radiol 2001; 11:2444–2448.

28. Feczko JD, Lynch L, Pless JE, et al. An autopsy case re-view of 142 nonpenetrating (blunt) injuries of the aorta. JTrauma 1992; 33:846–849.

29. Shkrum MJ, McClafferty KJ, Green RN, Nowak ES, YoungJG. Mechanisms of aortic injury in fatalities occurring in mo-tor vehicle collisions. J Forensic Sci 1999; 44:44–56.

30. Fujikawa T, Yukioka T, Ishimaru S, et al. Endovascularstent grafting for the treatment of blunt thoracic aortic injury.J Trauma 2001; 50:223–229.

31. Karmy-Jones R, Hoffer E, Meissner MH, Nicholls S, MattosM. Endovascular stent grafts and aortic rupture: a case se-ries. J Trauma 2003; 55:805–810.

32. Kearney PA, Smith DW, Johnson SB, et al. Use of trans-esophageal echocardiography in the evaluation of traumaticaortic injury. J Trauma 1993; 34:696–701.

33. Shiau YW, Wong YC, Ng CJ, Chen JC, Chiu TF. Periaorticcontrast medium extravasation on chest CT in traumaticaortic injury: a sign for immediate thoracotomy. Am J EmergMed 2001; 19:229–231.

34. Coady MA, Rizzo JA, Elefteriades JA. Pathologic variantsof thoracic aortic dissections: penetrating atherosclerotic ul-cers and intramural hematomas. Cardiol Clin 1999; 17:637–657.

35. Kepros J, Angood P, Jaffe CC, Rabinovici R. Aortic intimalinjuries from blunt trauma: resolution profile in nonoperativemanagement. J Trauma 2002; 52:475–478.

36. Holmes JH, Bloch RD, Hall RA, et al. Natural history of trau-matic rupture of the thoracic aorta managed nonoperatively:a longitudinal analysis. Ann Thorac Surg 2002; 73:1149–1154.

37. Gavant ML. Helical CT grading of traumatic aortic injuries:impact on clinical guidelines for medical and surgical man-agement. Radiol Clin North Am 1999; 37:553–574.

38. Mirvis SE. Diagnostic imaging of acute thoracic injury.Semin Ultrasound CT MR 2004; 25:156–179.

NO

TES

91

CT of Abdominal Trauma:Part I1

Trauma is the leading cause of death in individuals younger than 5 years old andranks third or fourth as the cause of death in the whole population of the UnitedStates and western Europe (1). About 10% of all trauma deaths are due to abdominalinjuries. Both compressive and deceleration mechanisms are at work with blunttrauma. Compression may injure the solid organs or the vessels, resulting in lacera-tions, hematomas, and thrombosis, and may injure the hollow viscera, resulting inrupture. Deceleration forces cause stretching and shearing, especially of vessels be-tween the relatively fixed origin of the vessel and the relatively mobile viscera. Figures1–10 illustrate and exemplify the range of findings shown with computed tomogra-phy (CT) in patients with blunt-force trauma to the abdomen and pelvis.

EVOLUTION OF TRAUMA MANAGEMENT

In the early 1900s, mortality from trauma was noted to be high without surgery; andaggressive diagnosis, with laparotomy and surgical treatment when possible, was under-taken (2). In 1965, diagnostic peritoneal lavage was shown to be useful for depicting thepresence of large volumes of intraperitoneal fluid and for determining the type offluid (3). In the 1970s, ultrasonography (US) was applied to diagnosis in the patientwith multiple trauma and was useful in depicting not only large volumes of intraperi-toneal fluid but also certain visceral injuries. US was, and continues to be, limited inthe characterization of injuries, the depiction of small volumes of intraperitoneal fluid,the differentiation of blood from other fluid, and the depiction of specific injuries thatwould require immediate surgery. Exploratory laparotomy remained the mainstay ofdiagnosis in the patient with multiple trauma.

Problems with aggressive surgery became evident. In the early 1950s, mortality frominfection was noted in infants after splenectomy. Hepatic surgery following traumaticinjury was complicated by uncontrollable hemorrhage. Lengthy surgery had its owncomplications, especially when accompanied by loss of large volumes of blood. Therate of nontherapeutic laparotomy for trauma shows variation from one institution toanother, but the rate has been about 20%. Morbidity has been seen at a rate of 10%–20% in trauma patients after “negative” (nontherapeutic) laparotomy (4). Thus,there was a desire within the surgical community to select those patients for whomsurgery would be the indicated therapy if diagnosis could be improved (2).

James T. Rhea, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 91–100.

1From the Departments of Radiology, Harvard Medical School and Massachusetts General Hospital, FH 210, 55 Fruit St,Boston, MA 02114 (e-mail: rhea.james@mgh.harvard.edu).

J.T.R. holds stock in General Electric Co (purchased on open market).

Rhea

92

The use of CT in the patient with multiple traumabegan in the 1980s. With improved CT technologyand improved accuracy, CT has become the diagnosticmethod of choice, replacing exploratory laparotomy.In addition, the development of interventional radio-logic techniques of therapeutic embolization hasmade it possible, in patients who have certain intra-abdominal injuries, to treat active bleeding with lessmorbidity and mortality than was possible previously(4). Today, the indications for surgery include uncon-trollable hypotension, signs of peritonitis, and specificlife-threatening injuries. CT can be used to detect activearterial extravasation and the specific type and grade ofinjury, allowing the selective use of surgical or other in-terventional therapy. Incorporation of CT into the earlyevaluation of the patient with multiple trauma has al-lowed nonsurgical management and has been shownto decrease the mortality rate of these patients (5).

With nonsurgical management, patients must beclosely observed because the rate of failure of nonsur-gical management has been found to be 10%–20%,depending on the organ that has been injured (6,7).Predictors of failure of nonsurgical management havebeen defined, and many of these findings can be seenat CT. These findings include a higher grade of injuryto the spleen, active arterial extravasation, hypoten-sion at manifestation, and a large volume (>300 mL)of free blood within the abdomen (6).

Because of its accuracy, CT has facilitated the changeto nonsurgical management of many cases of multipletrauma. CT can be used to define specific injuries thatare indications for immediate intervention. In addi-tion, CT can be used to define those findings that, aspredictors of failure of nonsurgical management, re-quire closer observation of the patient.

CT TECHNIQUES

Evaluation of the Abdomen

To optimize the accuracy of CT, both orally and intra-venously administered contrast materials are widelyused. One dose of oral contrast material may consist of¼ oz (7.5 mL) of diatrizoate meglumine in 10 oz (300mL) of fluid. Usually, scanning is not delayed for com-plete opacification of the bowel, but the presence of oralcontrast material in even the proximal portion of thebowel allows better assessment of the duodenum andthe pancreas. Extravasation of oral contrast material, al-though not sensitive, would be 100% specific for bowelinjury. Oral contrast material may be given in thetrauma bay after placement of a nasogastric tube andagain just prior to scanning. If time permits, a full doseregimen of oral contrast material would consist of three10-oz (300-mL) portions given at 30-minute intervals.

Intravenous contrast material administration consistsof power injection of 135 mL of a nonionic agent thatis approximately 60% iodine at a rate of at least 2.5

mL/sec. The use of intravenous contrast material is nec-essary to identify debilitating vascular injuries and activearterial bleeding and to better characterize the type oforgan injuries that occur.

Both oral and intravenous contrast materials carry asmall risk. Aspiration of oral contrast material may oc-cur, although the frequency of aspiration is low. Investi-gators have found that between 1 in 500 and 1 in 1000patients will aspirate oral contrast material (8,9). The useof oral contrast material continues to be the subject of re-cent investigations, with mixed conclusions about thenecessity of its use (10). Intravenous contrast materialcarries the usual low risk of an adverse reaction to thecontrast material. The benefit of improved identificationof life-threatening injury would appear to outweigh therelatively low risk associated with contrast material.

CT scan parameters for four–detector row helical CTscanning include the following: (a) 75-second delay afterintravenous injection of 135 mL of nonionic contrastmaterial (300 mg of iodine per milliliter) at 2.5 mL/sec;(b) scan from above the dome of the diaphragm to sacralvertebra S1; (c) additional 120-second delay to allow en-hancement of the distal portion of the ureters and thebladder; (d) scan from S1 to the ischial tuberosities;(e) detector configuration, 4 × 1.25 mm; (f) 5-mm sec-tion thickness and image spacing; (g) table speed, 15mm/sec; and (h) standard algorithm.

CT scan parameters for 16–detector row helical CTscanning are undergoing revision to reduce dosage andcurrently include the following: (a) 75-second delay afterintravenous injection of 135 mL of nonionic contrastmaterial (300 mg of iodine per milliliter) at 3.0 mL/sec;(b) scan from above the dome of the diaphragm to verte-bra S1; (c) additional 120-second delay to allow en-hancement of the distal portion of the ureters and thebladder; (d) scan from S1 to the ischial tuberosities;(e) detector configuration, 16 × 1.25 mm; (f) 5-mm sec-tion thickness and image spacing; (g) table speed, 18.75mm/sec; (h) pitch, 0.938; (i) tube rotation time, 0.5 sec-ond; and (j) standard algorithm.

Prior to scanning of the abdomen, the arms of thepatient should be placed up near the head to minimizeartifacts, the Foley catheter should be clamped, and thecardiac leads should be removed, if possible, to mini-mize artifacts. After the initial complete scan, delayedscanning should be performed through any suspectedpancreatic, renal, or bowel injury. This delayed scan-ning allows better characterization of the injury. Inaddition, CT data should be reformatted for evaluationof the spine. Sagittal and coronal reformations of theentire abdomen are made with a detail algorithm fromthe primarily acquired 1.25-mm image data used toreconstruct an axial image set.

Evaluation of the Spine

One to two percent of the patients with blunttrauma have a spine fracture, including Chance and

CT o

f Ab

do

min

al Traum

a: Part I

93

burst fractures. Attention may be diverted from thespine because about half of the trauma patients withspine fractures have associated injuries, about half ofwhich are intraabdominal.

Investigators have shown that “screening” the lum-bar spine for injury is more accurate with use of thedata from the abdominal CT examination than withthe use of separate conventional radiographic imagingof the spine (11–13). To evaluate the spine, the 5-mmaxial CT images are reconstructed to at least contigu-ous 2.5-mm axial images. These thinner axial imagesare then used to make sagittal and coronal reforma-tions of the spine. Viewing the axial CT images andthe sagittal and coronal reformations with the bonewindow allows assessment of the spine. It is advisableto leave the field of view on the sagittal and coronalimages open to include the soft tissues of the abdo-men. At times, the diagnosis of vascular and visceralinjuries is improved if these injuries are viewed in thesagittal or coronal plane.

Evaluation of the Urinary Bladder: CT Cystogram

With the Foley catheter clamped, extravasation ofurine leads to the diagnosis of urinary bladder rupture.However, a full-appearing bladder without extravasa-tion does not exclude bladder rupture. To confidentlyexclude bladder rupture, CT cystography is necessary.

The technique for CT cystography is as follows:(a) Unclamp the Foley catheter, and drain the bladder.(b) Then instill 300–400 mL of bladder contrast mate-rial through the Foley catheter. This contrast materialconsists of 40 mL of diatrizoate meglumine in 1 L ofnormal saline. (c) Rescan the pelvis from above theiliac crests to the symphysis pubis.

MULTITRAUMA ABDOMINAL CT:WHAT TO LOOK FOR

Active Arterial Extravasation

Active arterial extravasation correlates with a failureof conservative management of the patient and is seenin 10%–20% of the patients with blunt trauma (14–17).The results of investigations have shown that about78% of the patients who demonstrate active arterialextravasation at CT will require surgery or therapeuticembolization to attain hemodynamic stability. Thiscontrasts with the fact that only 28% of the patientswithout extravasation require surgery (15,16). Thus, itis important to note and inform our clinical colleaguesof any active arterial extravasation that is seen.

Extravasation has been described as a “vascularblush.” The finding evident at CT is the presence of ex-travascular contrast material adjacent to or within theinjured structure. Delayed scanning may demonstratean increased volume of extravasation of contrast materi-al. The sources of active extravasation may frequently beidentified at CT and include the viscera, vessels, and

mesentery. The frequency of extravasation varies fromrates of 20%–24% for kidney, mesentery, and pelvicinjuries to less than 10% for liver and adrenal injuries.

Hypotension

Hypotension results in specific findings that can benoted at CT (Fig 3) (18–22). These findings includethe following: (a) flat inferior vena cava (3-cm-longsegment of the inferior vena cava with a transverse-to-anteroposterior diameter ≤3:1); (b) small aorta (<6-mm diameter below the superior mesenteric artery),especially in children; (c) hyperemic bowel mucosa;(d) bowel wall thickening; (e) bowel dilatation withfluid; (f) hyperemic kidneys and adrenal glands; and(g) hypoperfusion of the spleen.

The hyperemic bowel mucosa is due to two factors,which have been demonstrated experimentally. Withhypotension, blood flow shifts to the mucosa of thebowel, and the transit time of the blood through thebowel wall is prolonged. The small aorta and hypoper-fusion of the spleen are thought to be due to arterialconstriction; such constriction is more likely in childrenthan in adults with some degree of arteriosclerosis.

Free Fluid

Isolated free fluid may occur without any explana-tory injury being found. The frequency of isolated freefluid in patients with multiple trauma is 2%–3%(23,24). This isolated free fluid may be the result ofrapid fluid resuscitation, in which case the finding istotally benign, except for a possible contribution toabdominal compartment syndrome. However, othermore important causes of isolated free fluid are pos-sible, and the finding may present a diagnostic di-lemma. The options for management when isolatedfree fluid is seen include observation, diagnostic peri-toneal lavage, laparotomy, or repeat CT.

Free fluid is seen in about 75% of the patients withintraabdominal injury (25). If free fluid is seen in thepatient with multiple trauma, the location, type, andvolume of the fluid should be assessed. The locationof free fluid can be either intraperitoneal or extraperi-toneal. The type of fluid could be blood, urine, bowelcontent, bile, ascites, or diagnostic peritoneal lavagefluid. The volume may be characterized as minor,moderate, or major.

Location.—The intraperitoneal compartments in-clude the perisplenic and perihepatic spaces, theMorison pouch between the right kidney and the tipof the liver, the pericolic gutters, the inframesocolicspace, the lesser peritoneal sac, and the pelvis. In addi-tion, small amounts of free fluid may be seen as trian-gular collections between the leaves of the mesentery.

The extraperitoneal compartments include the ante-rior pararenal, perirenal, posterior pararenal, peri-vesicle, anterior prevesicle, and pericholecystic spaces.The anterior pararenal space contains the pancreas

Rhea

94

and duodenum. The perirenal space contains the kid-neys, adrenal glands, inferior vena cava, and aorta.The posterior pararenal, perivesicle, and anteriorprevesicle spaces contain fat.

The location of blood in a particular compartmentcorrelates with an injury to a structure within thatcompartment. In two areas, the location may appearconfusing at CT. One area is the space between theliver and kidney. Blood in both the intraperitonealspace (Morison pouch) and the retroperitoneal ante-rior pararenal space will appear between the liver andkidney. To differentiate which location is involved,you must determine whether the blood wraps aroundthe inferior tip of the liver. The anterior pararenal fas-cia tightly adheres to the lateroconal fascia, and bloodin the retroperitoneal anterior pararenal space will notwrap around the tip of the liver (Fig 1a). In contrast,blood in the Morison pouch will wrap around the in-ferior tip of the liver (Fig 1b).

The second area that may present problems in inter-pretation is the pelvic intraperitoneal space and thefat-containing extraperitoneal spaces around the uri-nary bladder. One such extraperitoneal space is theanterior prevesicle space (Fig 2b). This is a potentialspace extending deep to the rectus muscles along theanterior abdominal wall from the level of the urinarybladder to the umbilicus. The fat around the bladderis also continuous with the pelvic sidewalls and thepresacral spaces. The definitive finding of intraperito-neal urine or blood or contrast material is that thefluid surrounds bowel loops (Fig 2a).

Type.—The type of fluid is assessed by measuringthe attenuation of the fluid in Hounsfield units. Thefollowing CT attenuation values are characteristic:(a) blood, more than 25 HU, with unclotted bloodmeasuring 25–50 HU and clotted blood measuring40–75 HU; and (b) ascites, bile, urine, bowel con-tents, and lavage fluid, 5–10 HU.

A mixture of blood with other fluid will be of inter-mediate attenuation, and the cause of the fluid cannotbe specified. Because the finding of free bile, urine, orbowel contents is serious, peritoneal lavage shouldonly be performed after CT scanning, not prior toscanning. Peritoneal lavage after scanning may be use-ful in distinguishing the type of low-attenuation fluidthat is present.

Volume.—The volume of fluid is of less importancethan the hemodynamic status of the patient. However,a large volume of blood within the peritoneum ismore likely to lead to pyrexia and correlates with fail-ure of nonsurgical management if the volume isgreater than 300 mL (6). The volume may be esti-mated by the number of intraperitoneal compart-ments in which fluid is seen. Volume may be charac-terized as minor (100 mL to <200 mL) when fluidoccupies one compartment. Moderate volume (200–500 mL) tends to occupy two compartments. Majorvolume (>500 mL) tends to be seen in three or morecompartments.

A small amount of fluid may be important andmay be the only indication of bowel or other injury.CT has been shown to depict as little as 10 mL of freefluid (26). US is not reliable in depicting smallamounts of fluid. The results of experimental workwith cadavers have shown that a volume of 100 mL isnecessary for US to reliably depict free fluid (27).

Associated Injuries

Multiple injuries tend to occur together. Focusingon a dramatic injury may lead to overlooking otherless obvious injuries. Two phrases are useful as re-minders to keep looking until a thorough search ofthe images has been completed: the “seat-belt syn-drome“ and “packages of injury.“

To our knowledge, the seat-belt syndrome was firstdescribed by Garrett and Braunstein (28) in 1962, and

Figure 1. Intraperitoneal versusextraperitoneal blood. (a) AxialCT image shows blood betweenthe liver and kidney. Note thatblood does not wrap around thetip of the liver (arrow) and is inthe anterior pararenal spacefrom a laceration of the barearea of the liver. (b) In contrast,this axial CT image shows thatthe blood in this patient wrapsaround the tip of the liver (ar-rows) and is in the peritonealspace from a spleen laceration.

CT o

f Ab

do

min

al Traum

a: Part I

95

the term refers to the simultaneous occurrence of bruis-ing of the abdominal wall, bowel or mesenteric injury,solid organ injury, and spine fracture. Major vesselsmay also be injured, as the term seat-belt aorta indicates(29,30). The mechanism of the seat-belt syndrome isthought to be compression of intraabdominal struc-tures between the spine and the anterior abdominalwall by a lap belt during a motor vehicle collision.

The packages of injury include the left, middle, andright packages (31). An injury in a given region islikely to be accompanied by other injuries nearby. Theleft package would include combinations of simulta-neous injuries to the left lobe of the liver, the spleen,the left kidney or adrenal gland, the distal portion ofthe pancreas, and/or left-sided thoracic injuries. Theright package would include similar injuries on theright. The midline injuries include combinations ofinjuries to the liver, transverse colon, small bowel,mesentery, pancreas anterior to the spine, duodenum,

aorta, inferior vena cava, and/or spine and midlinethoracic injuries.

INJURIES TO SPECIFIC ORGANS

Gallbladder Injury

Injury to the gallbladder occurs in 2%–8% of the pa-tients with blunt abdominal trauma (32). Associatedinjuries are frequent, with liver injury occurring inabout 80% and duodenal injury in about 50% of suchpatients. Injury to the extrahepatic bile duct occursabout one-fourth as often as gallbladder injury. Rupturemay be extraperitoneal, around the gallbladder fossa, ormay be intraperitoneal. If the rupture is intraperitoneal,a bile-induced peritonitis may occur. If the rupture isextraperitoneal, peritoneal signs will be absent, andpericholecystic fluid will be seen at CT (Fig 4).

The types of gallbladder injury seen include wallcontusion, rupture, and avulsion of the gallbladder

Figure 2. Intraperitoneal versus extraperitoneal contrast material. (a) Axial CT image shows contrast material from a bladder rup-ture surrounding bowel loops (arrows). This indicates that there has been an intraperitoneal rupture. (b) This axial CT image showsthat contrast material (arrows) from a bladder rupture is anterior to the bowel and deep to the rectus muscles. This indicates that therupture is extraperitoneal, with contrast material in the anterior prevesicle space.

Figure 3. Axial CT imagesdemonstrating signs of hypoten-sion. (a) CT image shows hyper-emic bowel mucosa (black arrow)and a narrow inferior vena cava(white arrow), which are indica-tions of hypotension. (b) Hypo-tension may also result in hypo-perfusion of the spleen (blackarrow) and a hyperattenuatingkidney (white arrow).

Rhea

96

from its fossa. CT findings to look for with gallbladderinjury include the following: (a) irregular ill-definedwall contour, (b) wall thickening, (c) wall discontinu-ity, (d) intraluminal (high-attenuation) blood, (e) col-lapsed lumen, (f) mucosal flap, and (g) blood or low-attenuation bile adjacent to the gallbladder.

Pancreatic Injury

Injury to the pancreas has been reported in fromless than 1% to 3% of the patients with multipletrauma and most often results from compression ofthe pancreas against the spine (32,33). Associated in-juries occur in about 70% of the adult cases but only15%–30% of the pediatric cases, with duodenal andliver injuries being the most frequent. Early diagnosisis important because delay in diagnosis correlateswith increased morbidity (34). Unfortunately, the di-agnostic accuracy of CT has not been as great in acutepancreatic injury as in most other injuries. Recent ad-vances in technology offer possible improvement, butthe findings from earlier investigations indicated thatthe sensitivity of CT for diagnosing pancreatic injurywas 68%–80% (35,36). The difficulty in diagnosis iscompounded by the fact that the serum amylase levelis neither sensitive nor specific for pancreatic injury inthe acute setting (34). However, 1–2 days after injury,the amylase level will be increased in 80%–90% ofthe patients with pancreatic injury.

Pancreatic injuries include contusion (Fig 5), lacera-tion, and fracture (Fig 6). Contusion may be focal ordiffuse and may be seen as a hypoattenuating or iso-attenuating area within the pancreas. One clue foridentification of isoattenuating contusion is asymme-

try of the pancreatic septa. The appearance of the septais variable from patient to patient. Younger patientsmay have no apparent septa, and older patients mayhave prominent septa. If septa are seen, look for anarea in which the septa are less evident than elsewherewithin the pancreas.

Lacerations are more easily recognizable, except forpossible confusion with a normal cleft that may occurin the anterior aspect of the pancreatic body anterior tothe spine. Lacerations are usually perpendicular to the

Figure 6. Pancreatic fracture. Axial CT image shows that thehead of the pancreas (white arrow) is separated from the body(black arrow) by a pancreatic fracture. Abundant blood is seenaround the pancreas, superior mesenteric vein, and superiormesenteric artery. The pancreatic duct, though not seen, wouldbe torn when a fracture or laceration crosses more than 50% ofthe width of the pancreas.

Figure 5. Pancreatic contusion. Axial CT image shows anarea in the tail of the pancreas (arrows) in which the subtlesepta seen elsewhere are absent. This represents an area ofpancreatic contusion. The patient had transient elevation of theserum amylase level.

Figure 4. Gallbladder rupture. Axial CT image of patient withrupture of the gallbladder shows blood in the pericholecysticspace and areas of active arterial extravasation (arrows). Be-tween the arrows, the wall of the gallbladder is not depicted,which is consistent with rupture.

CT o

f Ab

do

min

al Traum

a: Part I

97

axis of the pancreas and most frequently occur in theneck. Fractures of the pancreas do not present muchdiagnostic difficulty because of the separation of thefragments and the abundant blood in the anteriorpararenal space.

Injury to the pancreatic duct occurs in about 15%of the patients with pancreatic injury (33) and is notdirectly seen during the acutely performed abdominalCT examination. Pancreatic duct injury may be im-plied if a laceration is larger than one-half the thick-ness of the pancreas. Duct injury is a surgical emer-gency, and if there is evidence of injury in the anteriorpararenal space, other methods of evaluating the duct,such as endoscopic retrograde cholangiopancreatog-raphy or magnetic resonance cholangiopancreatogra-phy, may be considered (34,37,38).

Blood in the anterior pararenal space may be seen invarious locations. One place to look is between the pan-creas and the splenic vein. Normally, the splenic vein isclosely applied to the body of the pancreas without in-tervening fluid or fat. Blood may be located around thesuperior mesenteric artery, in the transverse mesocolon,or in the lesser sac and may appear as thickening of themargins of the left anterior pararenal fascia. If fluid isseen in the anterior pararenal space, always measure theattenuation to determine if the fluid is simple low-attenuation fluid rather than blood. If rapid resuscitativeadministration of fluids leads to intra- or extra-peritoneal accumulations, the most frequent locationof such fluid is in the anterior pararenal space.

If pancreatic injury is seen, grading the injury is use-ful because the difficulty of surgery increases with thegrade of injury. The classification of the American Asso-ciation for the Surgery of Trauma (AAST) (39) includesthe following classes: (a) class I, minor contusion, su-perficial laceration, intact duct; (b) class II, major con-tusion, major laceration, intact duct; (c) class III, ductinjury, distal transection; (d) class IV, proximal transec-tion, ampullary injury; and (e) class V, massive disrup-tion of pancreatic head.

In addition to the AAST classification, Lucas (40)developed the following classification that includesassociated injury to the duodenum: (a) class I, contu-sion, minor peripheral laceration, duct intact; (b) classII, deep laceration, transection of the body or tail, ductmay be damaged; (c) class III, severe injury to the headof the pancreas, duct may be damaged, duodenum in-tact; and (d) class IV, combined pancreatic and duode-nal injuries.

Adrenal Injury

Adrenal injury is seen in about 2% of the patientswith blunt abdominal trauma (41). Because the adre-nal glands are well protected in the retroperitoneum,other associated injuries are frequent in patients withadrenal injury from blunt abdominal trauma and areseen in more than 90% of these patients. The rightadrenal gland is injured more frequently than the left,and injury is usually unilateral. Bilateral adrenal in-jury may cause acute adrenal crisis, with hypotension,hypokalemia, hyponatremia, and acidosis (42).

CT findings of adrenal injury may be grouped into(a) findings that involve the adrenal gland itself and(b) secondary findings (43). The adrenal gland maycontain a round or oval mass, or hematoma (Fig 7);and this is the most common manifestation, seen inabout 80% of injured adrenal glands. Diffuse hemor-rhage will obscure the adrenal gland in about 10% ofinjured adrenal glands. A uniformly enlarged and in-distinct adrenal gland will be seen in about 10% of in-jured adrenal glands. Active arterial extravasation maybe demonstrated, and delayed hemorrhage may occur(44). Secondary findings of adrenal injury include fatstranding around the adrenal gland, thickening of thecrus of the diaphragm, thickening of the adjacent fas-cia, and frank retroperitoneal hemorrhage.

As seen on the trauma CT image, the shape and at-tenuation of an adrenal hematoma may be the sameas those of an adrenal neoplasm. If follow-up CT im-ages are obtained, the hematoma may increase in

Figure 7. Adrenal hematoma.(a) Axial CT image shows a rightadrenal hematoma with points ofactive arterial extravasation (ar-rows). (b) Two weeks later, re-peat axial CT image shows anincrease in size of the adrenalhematoma (arrows).

Rhea

98

size during the next few days and decrease in attenua-tion. A hematoma should have decreased in size at6 months after injury, and adrenal atrophy may beseen at 12 months.

Urinary Bladder Injury

Bladder rupture (Fig 8) is seen in 5%–10% of thetrauma patients with pelvic fracture and in about 10%of the trauma patients with gross hematuria (45,46).Rarely, bladder rupture may occur without hematuria.Associated injuries are seen in about 90% of suchbladder injuries because the bladder is well protectedby the bones of the pelvis. The mechanism of injury iseither pressure on a full bladder or penetration by abone fragment.

On the initial CT examination, a “full-appearingbladder” with the Foley catheter clamped does not ex-clude bladder rupture, and CT cystography is indicatedif there is a suspicion of bladder injury. Investigatorshave shown that the sensitivity of abdominal CT with-out cystography is only about 60% (47,48). The sensi-tivities of the retrograde fluoroscopic cystogram (85%–100%) and the CT cystogram (95%–100%) are similar(49–51). Details of the technique for CT cystographyare given in the section, “Evaluation of the UrinaryBladder: CT Cystogram.”

The types of bladder injury include contusion orinterstitial injury, in which hematoma or contrastmaterial is noted in the bladder wall, and rupture. Ifrupture is present, urine or contrast material will beseen in the intraperitoneal space, the extraperitonealspaces, or both locations. It is important to distinguishintraperitoneal from extraperitoneal rupture becausethe treatment is different. Intraperitoneal rupture,

which occurs in 10%–20% of the cases of bladderrupture, requires surgery. Extraperitoneal rupture,which occurs in about 80%–90% of the cases of blad-der rupture, requires only decompression with a Foleycatheter until healing takes place. Rupture into bothspaces occurs in 5%–10% of the patients with bladderrupture and also requires surgery.

To differentiate intraperitoneal from extraperitonealrupture of the bladder, look for urine or contrast ma-terial accumulating around bowel loops, in the perito-neal recesses, or between the leaves of the mesentery.Fluid between bowel loops in the mesentery will ap-pear as small triangular areas of soft-tissue density;bowel does not usually have this triangular appearance.The location of the extraperitoneal spaces around thebladder is noted in the section, “Free Fluid.”

Figure 8. Combined intraperitoneal and extraperitoneal bladder rupture. (a) Axial CT image through the bladder obtained before CTcystography shows the point of rupture of the bladder (black arrow) and unopacified urine, which is hypoattenuating, in the perivesiclespace anteriorly, which is continuous with the anterior prevesicle space (white arrows) seen in b. (b) Axial CT image obtained after CTcystography shows contrast material around a loop of bowel (arrow), consistent with the intraperitoneal component of the rupture, whichwas confirmed at surgery. The contrast material just deep to the rectus muscles is in the anterior prevesicle space.

Figure 9. Portal vein laceration. Axial CT image shows thatthe contour of the portal vein is irregular (arrow), and there isadjacent hemorrhage. Extrahepatic portal vein laceration wasfound at surgery.

CT o

f Ab

do

min

al Traum

a: Part I

99

Major Vascular Injury

The major vessels may be injured in the trauma pa-tient, and the CT appearance may be subtle. Associatedinjuries are frequent and can distract attention fromevaluation of the vascular structures. The findings tolook for with any vascular injury include the following:(a) irregular contour of vessel (Fig 9), (b) indistinctedge of vessel, (c) narrowing of lumen, (d) abrupt cut-off of vessel, (e) arterial pseudoaneurysm or flap,(f) extravasation of contrast material, and (g) hemor-rhage with vessel at the center (52–55).

Injury to the inferior vena cava may demonstrateadditional findings, including a liver laceration ex-tending into the porta hepatis and, rarely, fat herniat-ing into the lumen. The location of the injury to theinferior vena cava is important to specify because sur-gery becomes more difficult with the more proximalinjuries. Location may be specified as (a) infrarenal,(b) suprarenal and infrahepatic, (c) retrohepatic, or(d) suprahepatic.

The extrahepatic portion of the portal vein and thehepatic veins are subject to shearing injury withthrombosis. The dual blood supply of the liver mayresult in continued perfusion with portal vein injury.Thus, portal vein thrombosis may result in an absenceof depiction of the normally contrast material–enhanced intrahepatic portal veins but with paren-chymal enhancement via the hepatic artery. If thereis thrombosis of the hepatic veins, an absence ofperfusion with hypoattenuating segments of theliver may be seen because contrast-enhanced bloodcan enter the liver from neither the portal vein northe hepatic artery.

Injuries to the abdominal aorta are rare (56). Themost common site of aortic injury is near the inferiormesenteric artery, presumably related to a seat-belt

mechanism; the second most common site is near therenal arteries. The types of aortic injury that may beseen include partial thrombosis (Fig 10) or occlusion,intramural hematoma, pseudoaneurysm, late true an-eurysm, rupture, or intimal injury resulting in a flapor dissection (54). The major aortic branches mayalso be injured and should be carefully assessed.

References1. Novelline RA, Rhea JT, Bell T. Helical CT of abdominal

trauma. Radiol Clin North Am 1999; 37:591–612.2. Knudson MM, Maull KI. Nonoperative management of solid

organ injuries: past, present, and future. Surg Clin North Am1999; 79:1357–1371.

3. Fabian TC, Croce MA, Minard G, et al. Current issues intrauma. Curr Probl Surg 2002; 39:1160–1244.

4. Demetriades D, Velmahos G. Technology-driven triage of ab-dominal trauma: the emerging era of nonoperative manage-ment. Annu Rev Med 2003; 54:1–15.

5. Linsenmaier U, Krotz M, Hauser H, et al. Whole-body com-puted tomography in polytrauma: techniques and management.Eur Radiol 2002; 12:1728–1740.

6. Velmahos GC, Toutouzas KG, Radin R, Chan L,Demetriades D. Nonoperative treatment of blunt injury tosolid abdominal organs: a prospective study. Arch Surg2003; 138:844–851.

7. Toutouzas KG, Karaiskakis M, Kaminski A, Velmahos GC.Nonoperative management of blunt renal trauma: a prospec-tive study. Am Surg 2002; 68:1097–1103.

8. Nastanski F, Cohen A, Lush SP, DiStante A, Theuer CP. Therole of oral contrast administration immediately prior to thecomputed tomographic evaluation of the blunt trauma victim.Injury 2001; 32:545–549.

9. Federle MP, Peitzman A, Krugh J. Use of oral contrast materi-al in abdominal trauma CT scans: is it dangerous? J Trauma1995; 38:51–53.

10. Holmes JF, Offerman SR, Chang CH, et al. Performance ofhelical computed tomography without oral contrast for the de-tection of gastrointestinal injuries. Ann Emerg Med 2004; 43:120–128.

11. Rhea JT, Sheridan RL, Mullins ME, Novelline RA. Can chestand abdominal CT eliminate the need for plain films of thespine? Emerg Radiol 2001; 8:99–104.

Figure 10. Aortic injury withthrombosis. (a) Axial CT imagefrom patient who was in a motorvehicle collision shows that thelumen of the aorta is narrowed,with contrast material centrally(arrow) and thrombus peripher-ally. (b) Coronal reformation ofthe abdominal CT data showsthe short segment of almostcomplete thrombosis of theaorta (arrow). This patient alsohad a Chance fracture of the ad-jacent lumbar vertebra (notshown).

Rhea

100

12. Gestring ML, Gracias VH, Feliciano MA, et al. Evaluation ofthe lower spine after blunt trauma using abdominal computedtomographic scanning supplemented with lateral scanograms.J Trauma 2002; 53:9–14.

13. Sheridan R, Peralta R, Rhea J, Ptak T, Novelline R. Refor-matted visceral protocol helical computed tomographic scan-ning allows conventional radiographs of the thoracic and lum-bar spine to be eliminated in the evaluation of blunt traumapatients. J Trauma 2003; 55:665–669.

14. Wong YC, Wang LJ, See LC, Fang JF, Ng CJ, Chen CJ. Con-trast material extravasation on contrast-enhanced helicalcomputed tomographic scan of blunt abdominal trauma: itssignificance on the choice, time, and outcome of treatment. JTrauma 2003; 54:164–170.

15. Yao DC, Jeffrey RB Jr, Mirvis SE, et al. Using contrast-en-hanced helical CT to visualize arterial extravasation afterblunt abdominal trauma: incidence and organ distribution. AJRAm J Roentgenol 2002; 178:17–20.

16. Willmann JK, Roos JE, Platz A, et al. Multidetector CT: detec-tion of active hemorrhage in patients with blunt abdominaltrauma. AJR Am J Roentgenol 2002; 179:437–444.

17. Eubanks JW III, Meier DE, Hicks BA, Joglar J, Guzzetta PC.Significance of "blush" on computed tomography scan in chil-dren with liver injury. J Pediatr Surg 2003; 38:363–366.

18. Eisenstat RS, Whitford AC, Lane MJ, Katz DS. The "flat cava"sign revisited: what is its significance in patients withouttrauma? AJR Am J Roentgenol 2002; 178:21–25.

19. Jeffrey RB, Federle MP. The collapsed inferior vena cava: CTevidence of hypovolemia. AJR Am J Roentgenol 1988; 150:431–432.

20. Taylor GA, Fallat ME, Eichelberger MR. Hypovolemic shock inchildren: abdominal CT manifestations. Radiology 1987; 164:479–481.

21. Sivit CJ, Taylor GA, Bulas DI, Kushner DC, Potter BM,Eichelberger MR. Posttraumatic shock in children: CT findingsassociated with hemodynamic instability. Radiology 1992;182:723–726.

22. Mirvis SE, Shanmuganathan K, Erb R. Diffuse small-bowel is-chemia in hypotensive adults after blunt trauma (shockbowel): CT findings and clinical significance. AJR Am JRoentgenol 1994; 163:1375–1379.

23. Rodriguez C, Barone JE, Wilbanks TO, Rha CK, Miller K. Iso-lated free fluid on computed tomographic scan in blunt ab-dominal trauma: a systematic review of incidence and man-agement. J Trauma 2002; 53:79–85.

24. Ng AK, Simons RK, Torreggiani WC, Ho SG, Kirkpatrick AW,Brown DR. Intra-abdominal free fluid without solid organ injuryin blunt abdominal trauma: an indication for laparotomy. JTrauma 2002; 52:1134–1140.

25. Hahn DD, Offerman SR, Holmes JF. Clinical importance of in-traperitoneal fluid in patients with blunt intra-abdominal injury.Am J Emerg Med 2002; 20:595–600.

26. Miller MT, Pasquale MD, Bromberg WJ, Wasser TE, Cox J.Not so FAST. J Trauma 2003; 54:52–59.

27. Bennett MK, Jehle D. Ultrasonography in blunt abdominaltrauma. Emerg Med Clin North Am 1997; 15:763–787.

28. Garrett JW, Braunstein PW. The seat belt syndrome. JTrauma 1962; 2:220–238.

29. Dajee H, Richardson IW, Iype MO. Seat belt aorta: acute dis-section and thrombosis of the abdominal aorta. Surgery 1979;85:263–267.

30. Fontaine AB, Nicholls SC, Borsa JJ, Hoffer E, Bloch RD,Kohler T. Seat belt aorta: endovascular management with astent-graft. J Endovasc Ther 2001; 8:83–86.

31. Novelline RA, Rhea JT, Bell T. Helical CT of abdominaltrauma. Radiol Clin North Am 1999; 37:591–612.

32. Shuman WP. CT of blunt abdominal trauma in adults. Radiol-ogy 1997; 205:297–306.

33. Patel SV, Spencer JA, el-Hasani S, Sheridan MB. Imaging ofpancreatic trauma. Br J Radiol 1998; 71:985–990.

34. Cirillo RL Jr, Koniaris LG. Detecting blunt pancreatic injuries.J Gastrointest Surg 2002; 6:587–598.

35. Ilahi O, Bochicchio GV, Scalea TM. Efficacy of computed to-mography in the diagnosis of pancreatic injury in adult blunttrauma patients: a single-institutional study. Am Surg 2002;68:704–707.

36. Wales PW, Shuckett B, Kim PC. Long-term outcome afternonoperative management of complete traumatic pancreatictransection in children. J Pediatr Surg 2001; 36:823–827.

37. Canty TG Sr, Weinman D. Management of major pancreaticduct injuries in children. J Trauma 2001; 50:1001–1007.

38. Dondelinger RF, Boverie JH, Cornet O. Diagnosis of pancre-atic injury: a need to improve performance. JBR-BTR 2000;83:160–166.

39. American Association for the Surgery of Trauma. Injury scor-ing tables: Table 10—pancreas injury scale. Available at:www.aast.org. Accessed May 19, 2004.

40. Lucas CE. Diagnosis and treatment of pancreatic and duode-nal injury. Surg Clin North Am 1977; 57:49–65.

41. Iuchtman M, Breitgand A. Traumatic adrenal hemorrhage inchildren: an indicator of visceral injury. Pediatr Surg Int 2000;16:586–588.

42. Udobi KF, Childs EW. Adrenal crisis after traumatic bilateraladrenal hemorrhage. J Trauma 2001; 51:597–600.

43. Burks DW, Mirvis SE, Shanmuganathan K. Acute adrenal in-jury after blunt abdominal trauma: CT findings. AJR Am JRoentgenol 1992; 158:503–507.

44. Oto A, Ozgen B, Akhan O, Besim A. Delayed posttraumaticadrenal hematoma. Eur Radiol 2000; 10:903–905.

45. Morey AF, Iverson AJ, Swan A, et al. Bladder rupture afterblunt trauma: guidelines for diagnostic imaging. J Trauma2001; 51:683–686.

46. Morgan DE, Nallamala LK, Kenney PJ, Mayo MS, Rue LW III.CT cystography: radiographic and clinical predictors of blad-der rupture. AJR Am J Roentgenol 2000; 174:89–95.

47. Haas CA, Brown SL, Spirnak JP. Limitations of routine spiralcomputerized tomography in the evaluation of bladder trauma.J Urol 1999; 162:51–52.

48. Hsieh C, Chen R, Fang J, et al. Diagnosis and managementof bladder injury by trauma surgeons. Am J Surg 2002; 184:143–147.

49. Peng MY, Parisky YR, Cornwell EE III, Radin R, Bragin S. CTcystography versus conventional cystography in evaluation ofbladder injury. AJR Am J Roentgenol 1999; 173:1269–1272.

50. Deck AJ, Shaves S, Talner L, Porter JR. Computerized to-mography cystography for the diagnosis of traumatic bladderrupture. J Urol 2000; 164:43–46.

51. Vaccaro JP, Brody JM. CT cystography in the evaluation ofmajor bladder trauma. RadioGraphics 2000; 20:1373–1381.

52. Hewett JJ, Freed KS, Sheafor DH, Vaslef SN, Kliewer MA.The spectrum of abdominal venous CT findings in blunttrauma. AJR Am J Roentgenol 2001; 176:955–958.

53. Kimoto T, Kohno H, Uchida M, et al. Inferior vena cavalthrombosis after traumatic liver injury. HPB Surg 1998; 11:111–116.

54. Berthet JP, Marty-Ane CH, Veerapen R, Picard E, Mary H,Alric P. Dissection of the abdominal aorta in blunt trauma: en-dovascular or conventional surgical management? J VascSurg 2003; 38:997–1003.

55. Lin PH, Barr V, Bush RL, Velez DA, Lumsden AB, Ricketts J.Isolated abdominal aortic rupture in a child due to all-terrainvehicle accident: a case report. Vasc Endovascular Surg2003; 37:289–292.

56. Safriel YI, Sclafani SJ, Kurtz RS. Preoperative diagnosis ofright hepatic vein injury by CT scan and venography. JTrauma 2001; 51:149–152.

101

CT of Abdominal Trauma:Part II1

Diagnosis and treatment of patients admitted to a trauma center with potential bluntabdominal injury has been a difficult and challenging task for the trauma surgeon andemergency radiologist (1–4). Computed tomography (CT) is the imaging modality ofchoice to evaluate hemodynamically stable patients who have sustained blunt abdomi-nal trauma. During the past 5 years, single–detector row helical CT has been replaced bymulti–detector row CT. This change has revolutionized cross-sectional imaging intrauma radiology. Volumetric imaging with helical CT has been a major factor for non-surgical management of solid-organ injuries. The ability to obtain high-resolutionimages during optimal contrast enhancement at unparalleled speed has made multi–detector row CT the primary imaging modality of choice in evaluating hemodynami-cally stable patients with abdominal pain, tenderness, or positive ultrasonographic find-ings for free intraperitoneal fluid. Currently, a 16–detector row CT scanner can image theneck, chest, abdomen, and pelvis (from the circle of Willis to the symphysis pubis) inless than 60 seconds (5). This chapter will discuss the role of multi–detector row CT inthe diagnosis and management of splenic, liver, renal, bowel, and mesenteric injuries.

TECHNIQUE

At our trauma center, multi–detector row CT examinations of the abdomen and pel-vis of blunt trauma patients are performed (a) as part of a “whole-body” CT exami-nation, which includes the neck, chest, abdomen, and pelvis, or (b) from the lowerportion of the chest to the symphysis pubis. Contrast material (150 mL containing300 mg of iodine per milliliter) is routinely administered intravenously with a powerinjector (biphasic injection: 90 mL at 6 mL/sec, followed by 60 mL at 4 mL/sec) un-less there is (a) a known history of a major allergic reaction to iodinated contrastmaterial or (b) renal insufficiency. Delayed images are obtained routinely about 2–3minutes following intravenous injection of the contrast material to evaluate the re-nal collecting system for injuries. The CT parameters used at our institution forsingle–detector row helical and multi–detector row CT and for intravenous adminis-tration of contrast material are shown in the Table. Currently, studies are under wayat our institution to optimize the whole-body multi–detector row CT scanning pa-rameters and the rate, concentration, and volume of intravenous contrast material toobtain high-resolution images at peak contrast enhancement.

Kathirkamanathan Shanmuganathan, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 101–112.

1From the Department of Radiology, University of Maryland School of Medicine, 22 S Greene St, Baltimore, MD 21201(e-mail: kshanmuganathan@umm.edu).

Shan

mu

gan

ath

an

102

CT Protocols

Intravenous Injection TableType of Contrast Delay Rate Collimation Speed

Spiral CT Material (mL) (sec)* (mL/sec) (mm) (mm) Pitch

Single–detector row CT 150.0 60 3 8. 8 1.0Sixteen–detector row CT 90.0 70 6 16 × 1.5† 37.4 1.2

60.0 4Four–detector row CT 90.0 70 6 4 × 2.5† 10 1.0

60.0 4

* Delay in initiation of scan from beginning of intravenous bolus.† Detector configuration: number of sections (detectors) × section thickness (in millimeters). Number of sections is number

obtained per rotation (0.75 or 0.8 second).

Figure 1. Splenic lacerations and active bleeding. Multi–detector row axial CT im-ages obtained during (a) portal venous phase and (b) renal excretory phase showareas of active bleeding (arrowheads) and lacerations (straight arrows) seen within agrade V splenic injury. A large perisplenic hematoma (curved arrows) is also seen. Delayed image shows the active bleeding in-creases in extent. (c) Maximum intensity projection coronal image obtained in portal venous phase demonstrates active bleeding(arrowheads) and hematoma and hemoperitoneum (curved arrows) displacing the peritoneal content inferiorly and to the right side.

Although oral administration of contrast materialremains controversial in trauma centers, we routinelyadminister a total volume of 600 mL of 2% sodiumdiatrizoate orally or through a naso- or orogastric tubeat 30 minutes before and immediately before the scan.Patients who require urgent CT scanning on arrival atthe admitting area may be scanned immediately fol-lowing oral administration of one dose of contrastmaterial. Rectal contrast material is not routinely usedto evaluate blunt trauma patients.

SPLENIC INJURY

The spleen is the solid abdominal organ most com-monly injured with blunt trauma. During the past 2decades, CT has had an important effect in support-ing nonsurgical approaches to management of bluntsplenic trauma (6–8). Only recently has the vital role

played by the spleen in the immune defense systembeen fully appreciated, and this understanding has ledto a more conservative (nonsurgical) approach to themanagement of splenic injury, both in adults andchildren (6–11).

Splenic Injury CT Grading Systems

Although contrast material–enhanced CT is highlyaccurate in the diagnosis of splenic injury, CT gradingsystems have been generally unreliable in predictingoutcome following blunt splenic injury in adults (12–14). Many systems have been proposed to grade splenicinjury following trauma. The grades of splenic injurymay be based on the extent of injury seen at laparo-tomy, CT, or autopsy (12,14). None of the current sur-gical or CT-based splenic injury grading systems haveincorporated select predictive CT findings commonlyseen on contrast-enhanced studies, including active

CT o

f Ab

do

min

al Traum

a: Part II

103

splenic bleeding, pseudoaneurysms, or posttraumaticarteriovenous fistulas. In the recent radiologic and sur-gical literature, investigators have suggested that thesethree CT findings have a high association with failednonsurgical management (7,8,15,16).

CT Appearances of Splenic Injury

Contrast-enhanced CT can be used to accuratelydiagnose the four principal types of splenic injury:hematoma, laceration (Fig 1), active hemorrhage(Figs 1, 2) (17), and vascular injuries, includingpseudoaneurysm and posttraumatic arteriovenousfistula (Fig 3) (7,8,15,16,18). High-resolution over-lapping thin axial images generated with multi–detector row CT can be used by sophisticated multi-planar postprocessing programs to generate high-quality isotropic images. These images are used todetect and depict the relationship between the paren-chymal lesions and vascular structures and to differ-entiate the spectrum of splenic injuries seen follow-ing blunt trauma (Figs 1c, 2e, 3g).

Hematomas and laceration.—Splenic hematomasmay be intraparenchymal (Figs 2b, 3a) or subcapsular(Fig 2). Single or multiple hematomas may be seenfollowing blunt trauma. On contrast-enhanced CT im-ages, acute hematomas appear as irregular high- orlow-attenuation areas within the parenchyma. Oncontrast-enhanced CT images, subcapsular hemato-mas are typically seen as low-attenuation collectionsof blood between the splenic capsule and the enhanc-ing splenic parenchyma. On nonenhanced CT images,subcapsular hematoma is hyperattenuating relative tonormal splenic parenchyma. Subcapsular hematomasoften compress the underlying splenic parenchyma,and this CT finding helps to differentiate subcapsularhematomas from small amounts of blood or fluid inthe perisplenic space. Attenuation values of uncompli-cated subcapsular hematomas typically decrease withtime and resolve within 4–6 weeks.

Acute splenic lacerations have sharp or jagged mar-gins and appear as linear or branching low-attenua-tion areas on contrast-enhanced CT images (Fig 1b).With time, the margins of splenic lacerations and he-matomas become less well defined, and the lesionsdecrease in size until the area becomes isoattenuatingcompared with normal splenic parenchyma. Com-plete healing, as determined from the CT appearance,may take weeks to months, depending on the initialsize of the injury. Enlargement or development of anew lesion at follow-up CT should raise the possibil-ity of injury progression and warrants close clinicalobservation, further follow-up CT, or arteriography.

Active hemorrhage.—On contrast-enhanced CT im-ages, active hemorrhage in the spleen is seen as an ir-regular or linear area of contrast material extravasa-tion (Figs 1, 2). The difference between the attenua-tion value of extravasated contrast material (range,

85–350 HU; mean, 132 HU) and hematoma (range,40–70 HU; mean, 51 HU) is helpful in distinguishingactive bleeding from clotted blood (18,19). Activesplenic hemorrhage may be seen within the splenicparenchyma (Fig 2), subcapsular space (Fig 2), or in-traperitoneally (Fig 1b). On multi–detector row CTimages, ongoing hemorrhage may be seen as an in-crease over time in the amount of extravasation of in-travenous contrast material in the identical anatomicregion by comparing the arterial and delayed renal ex-cretory phases of the CT examination (Figs 1, 2).

Splenic vascular injuries.—The appearances of post-traumatic splenic pseudoaneurysms (Fig 3) and arte-riovenous fistulas are similar on contrast-enhancedmulti–detector row CT images; these lesions can onlybe differentiated with splenic angiography (8). Onmulti–detector row CT images, both of these lesionsappear as well-circumscribed focal areas of increasedattenuation compared with the normal enhancedsplenic parenchyma (attenuation within extravasa-tions typically measures within 10 HU of the attenua-tion of an adjacent major artery) (8,18). On imagesobtained in the delayed renal excretory phase, theselesions become minimally hyperattenuating orisoattenuating compared with the normal splenic pa-renchyma (Fig 3) (5).

With multi–detector row CT, the ability to scan dur-ing peak contrast enhancement in the early arterial orportal venous phase and during the excretory phaseaids in differentiating active bleeding from posttrau-matic splenic pseudoaneurysms or arteriovenous fistu-las (Figs 1–3). Usually, posttraumatic vascular injuriesare similar in attenuation value to active hemorrhage inthe arterial phase but “wash out” in the excretory phaseto become minimally hyperattenuating or isoatten-uating compared with normal splenic parenchyma (Fig3). Good correlation between the multi–detector rowCT findings of active splenic hemorrhage and the needfor angiographic or surgical intervention to treat hem-orrhage has led to an aggressive diagnostic and thera-peutic pursuit of splenic vascular injury with splenicangiography at our trauma center. Patients with CT evi-dence of splenic vascular contrast material extravasa-tion and without vascular injury at splenic arteriogra-phy undergo prophylactic proximal main splenic arteryembolization, which potentially increases the numberof patients with blunt splenic injuries who can betreated successfully without surgery.

Posttraumatic splenic infarction.—Segmental splenicinfarction is a rare manifestation of blunt trauma tothe spleen (20). On contrast-enhanced CT images,posttraumatic splenic infarcts are seen as well-demar-cated segmental wedge-shaped low-attenuation areas,with the base of the wedge toward the periphery ofthe splenic parenchyma (Fig 4). These infarcts can bethe only CT finding of blunt splenic trauma and mayoccur without any adjacent free fluid. Splenic infarcts

Shan

mu

gan

ath

an

104

Figure 2. Active bleeding from grade IVsplenic injury in a patient admitted follow-ing motor vehicle collision. (a, b) Multi–detector row axial CT images obtained inportal venous phase show a well-definedrounded pseudoaneurysm (black arrow-head) and linear areas of active bleedingwithin splenic parenchymal hematoma(white arrowhead) and subcapsular space(curved arrow). Active bleeding andpseudoaneurysms were similar in attenu-ation to intravenous contrast materialseen in splenic vessels (not shown).Large hematoma (straight arrows) is seenaround spleen. (c, d) Delayed multi–detector row axial CT images obtained insame region during renal excretoryphase show washout of contrast materialin splenic pseudoaneurysm and increasein parenchymal (arrowhead) and sub-capsular (curved arrow) hemorrhage. (e) Multiplanar curved coronal reforma-tion shows splenic injury (white arrowheads), subcapsular active bleeding (curved arrow), perisplenic clot (straight arrows), andhemoperitoneum (black arrowheads) adjacent to liver. (f) Early splenic arteriogram shows an actively bleeding pseudoaneurysm(arrow). (g) Delayed splenic arteriogram confirms intraparenchymal bleeding (straight arrow) and subcapsular bleeding (curvedarrows) seen on multi–detector row CT images. (Reprinted, with permission, from reference 17.)

CT o

f Ab

do

min

al Traum

a: Part II

105

Figure 3. Multiple posttraumatic splenicpseudoaneurysms. (a–c) Multi–detector rowaxial CT images obtained in portal venousphase show multiple splenic pseudoaneu-rysms (curved arrows) with hemoperito-neum (black arrowheads). Splenic lacera-tions (straight arrow) and a parenchymalhematoma (white arrowheads) are seen.(d–f) Delayed axial images obtained insame region during renal excretory phase.Pseudoaneurysms (arrow) wash out andbecome minimally hyperattenuating or iso-attenuating compared with normal splenicparenchyma. (g) Coronal maximum intensityprojection image shows three pseudoaneu-rysms (arrowheads). (h) Splenic arteriogramshows three pseudoaneurysms (arrow-heads), which were embolized. (Reprinted,with permission, from reference 17.)

Shan

mu

gan

ath

an

106

may also be seen in association with splenic lacera-tions and segmental infarcts in the kidney. Injury tothe intima of splenic artery branches caused by sud-den deceleration at the time of impact can lead tothrombosis and infarction of the splenic parenchymabecause of a lack of parenchymal perfusion distal tothe intimal injury. Similar injuries have been observedat CT of the kidneys following blunt trauma (5,20).Although the exact natural history of this injury is notknown, most of these lesions usually heal withoutneed for surgical or angiographic intervention. De-layed complications following posttraumatic splenicinfarction are rare and include splenic abscess forma-tion or delayed rupture of the spleen (20).

LIVER INJURY

The liver is the second most commonly injured solidorgan following blunt trauma. From 70% to 90% ofhepatic injuries are minor and either do not requiresurgery or have stopped bleeding at the time ofceliotomy (21,22). Because the right lobe constitutes80% of the hepatic volume, it is the most frequentlyinjured region. Mortality rates may exceed 50% inpatients with complex liver injuries who present withhemodynamic instability caused by active bleeding(23–26). Isolated injuries of the liver occur in lessthan 50% of the patients with blunt liver injury.

Hepatic Injury Grading System

Radiologic and surgical injury severity scores arebased on the anatomic disruption of the liver, includ-ing the depth and number of lacerations and the sur-face area involved by subcapsular or intraparenchymalhematomas seen at laparotomy (12,27,28). These in-jury scales have helped to standardize reporting of liverinjuries over a period of time and to compare out-comes and treatment protocols within the same traumacenter or among different centers.

Multi–Detector Row CT Appearance of LiverInjury

The four principal types of parenchymal liver injuryshown with CT are hematoma, laceration, vascular in-jury, and active hemorrhage. The results of numerousstudies have shown that CT can be used to accuratelydiagnose these four types of injuries and to guide careof blunt trauma patients with liver injury (27–30).

Hematoma.—Hepatic hematomas may be intrapar-enchymal (Fig 5) or subcapsular. On contrast-enhanced CT images, most subcapsular hematomasare seen along the anterolateral aspect of the rightlobe of the liver as a low-attenuation lens-shapedcollection of blood between Glisson capsule and theenhancing liver parenchyma. Subcapsular hemato-mas cause direct compression of underlying liver pa-renchyma, and this CT sign is helpful in differentiat-ing subcapsular hematoma from small amounts of

Figure 5. Intra-parenchymal he-patic hematoma.Multi–detector rowCT axial imageobtained on ad-mission showslarge intrapar-enchymal he-matoma (arrows)in the right lobe,with perihepaticblood (arrow-head).

Figure 4. Posttraumatic splenic infarction.(a) Axial multi–detector row CT image and(b) coronal multiplanar reformation showwedge-shaped low-attenuation area (ar-rows) of splenic infarction. No perisplenicfluid is seen.

CT o

f Ab

do

min

al Traum

a: Part II

107

free intraperitoneal blood or fluid seen adjacent tothe liver (perihepatic spaces). Follow-up CT imagestypically show resolution of uncomplicated subcap-sular hematomas within 6–8 weeks (31).

Contusion.—Parenchymal contusions of the liver ap-pear as irregular areas of low attenuation on contrast-enhanced CT images, with possible intermixed high-attenuation blood. On nonenhanced CT images, acutehematomas appear as irregular high-attenuation re-gions of clotted blood surrounded by lower-attenua-tion nonclotted blood or bile. With early healing, thecontusion may initially expand in size slightly and de-velop smoother, more regular margins, which shouldnot be taken as a sign of worsening injury. With fur-ther resolution, the lesion demonstrates a gradual de-crease in the attenuation and size of the hematomauntil there is blending with the background paren-chyma.

Laceration.—On contrast-enhanced multi–detectorrow CT images, liver lacerations appear as irregularlinear or branching low-attenuation areas (Fig 6). Thelocation of a laceration and its relationship to the he-patic veins may be important in predicting the likeli-hood of hemorrhage. Poletti et al (29) reported high-grade liver lacerations (grades III to V) involving theportal veins (Fig 6b) and the region of the majorbranches of the hepatic veins (Fig 6c); with activebleeding, such lacerations have an increased likeli-hood for major vascular injury and should prompthepatic arteriography in patients with clinical signs ofhemorrhage. Prior knowledge of lacerations extendinginto the region of the intrahepatic inferior vena cavaor the three major hepatic veins indicates a high likeli-hood of inferior vena cava or hepatic vein disruption.Hepatic lacerations with a branching pattern couldmimic the appearance of unopacified portal or he-

patic veins or dilated bile ducts and may require care-ful evaluation of serial images to differentiate amongthese various structures.

Acute liver lacerations have sharp or jagged margins.As the lesion heals, it enlarges, its margins becomesmoother, and it assumes a round to oval configurationon follow-up CT images. These lesions may graduallydecrease in size with time or remain as well-defined he-patic cysts.

Active hemorrhage.—Multi–detector row CT can of-ten be used to distinguish a pseudoaneurysm fromintraparenchymal extravasation by comparing the ap-pearance of the lesion during peak arterial enhance-ment and delayed “washout” imaging (delayed renalcollecting system phase). Pseudoaneurysms will showa more or less complete washout of contrast material,whereas parenchymal contrast enhancement from lo-cal tissue extravasation will persist and may be seen toincrease on the delayed images.

Liver lesions that could mimic active bleeding at CT,other than retained extravasated arteriographic con-trast material, include contrast enhancement seen inhepatic hemangiomas or other vascular tumors. Othersigns of hepatic trauma essentially always accompanybleeding of traumatic origin, but trauma-induced hem-orrhage from hypervascular hepatic tumors should alsobe considered.

Vascular injuries.—Retrohepatic vena cava injuries aretypically associated with high mortality (90%–100%)(23–25,32,33). Retrohepatic vena cava injuries are sus-pected at CT when (a) liver lacerations extend to themajor hepatic veins or inferior vena cava or (b) profusehemorrhage is present behind the right lobe of theliver, extends into the lesser sac, or collects adjacentto the diaphragm. On contrast-enhanced CT images,devascularized segments may appear as wedge-shaped

Figure 6. Hepatic lacerations extending into hepatic vein and porta hepatis. (a) Axialmulti–detector row CT image and (b, c) coronal maximum intensity projection imagesshow hepatic lacerations (straight arrows) extending into region of hepatic vein (curvedblack arrow) and porta hepatis (curved white arrow). Hemoperitoneum (arrowheads) isalso seen around liver.

Shan

mu

gan

ath

an

108

regions extending to the periphery that fail to enhancewith the normally perfused liver.

Pseudoaneurysms or injury of the hepatic artery andits branches may result from lacerations extendingacross the course of these vessels, as well as fromshearing forces. Arteriovenous or arterioportal fistulasmay mimic the contrast-enhanced CT appearance ofhepatic pseudoaneurysms. The findings from a retro-spective review of patients sustaining blunt hepaticinjury who had undergone both contrast-enhancedsingle–detector row spiral CT and hepatic arteri-ography showed that single–detector row contrast-enhanced helical CT had a sensitivity of 65% and aspecificity of 85% for detecting hepatic arterial injuryfor all CT injury grades when arteriography was thereference standard (29).

Periportal low attenuation.—Periportal low attenua-tion refers to regions of low attenuation that parallelthe portal vein and its branches at CT (Fig 7). This CTfinding has been noted in several nontraumatic clinicalconditions, including acute transplant rejection, malig-nant neoplasm of the liver, liver transplantation, car-diac failure, and cardiac tamponade and is usually at-tributed to dilatation of the intrahepatic lymph chan-nels caused by obstruction of the normal hepatic

lymphatic drainage. In trauma patients, periportalareas of low attenuation on CT images are often theconsequence of (a) vigorous intravenous fluid adminis-

Figure 7. Periportal low attenuation from vigorous volume re-suscitation. Multi–detector row CT image shows diffuse low-attenuation areas (arrowheads) paralleling portal vein and itsbranches, compatible with periportal edema. No liver injury wasseen. Inferior vena cava (curved arrow) is markedly distendedand larger than aorta from raised central venous pressure.

Figure 8. Full-thickness bowelinjury with subtle free intraperi-toneal air. (a, b) Multi–detectorrow axial CT images show freeintraperitoneal fluid (arrow-heads) in pelvis and upper por-tion of abdomen adjacent tospleen and liver. (c, d) Axial CTimages show proximal small-bowel wall thickening (black ar-rows), mesenteric infiltration (ar-rowheads), and bubble of freeintraperitoneal air (white arrow).At celiotomy, full-thickness jeju-nal and mesenteric injuries wererepaired. (Reprinted, with per-mission, from reference 17.)

CT o

f Ab

do

min

al Traum

a: Part II

109

tration prior to performing CT or (b) other trauma-re-lated causes of elevated central venous pressure (such astension pneumothorax or pericardial tamponade) thatresult in distention of the periportal lymphatic vessels(34,35). Periportal low attenuation seen on CT imageswithout evidence of parenchymal liver injury is not con-sidered to represent hepatic injury and does not warranthospitalization for observation or follow-up CT.

BOWEL AND MESENTERIC INJURIES

Bowel and mesenteric injuries are found in approxi-mately 5% of the patients sustaining blunt abdomi-nal trauma (36,37). The classic symptoms of rigidity,tenderness, and decreased or absent bowel soundsmay be present in only one-third of the patients (38).

In more recent studies with helical CT, investiga-tors have reported sensitivities ranging from 84% to94%, with accuracy from 84% to 99% in diagnosinghollow viscus injury (39,40). CT can be used to dif-ferentiate between surgical bowel injury (full-thick-ness bowel injury) and nonsurgical bowel injury (se-rosal tear or bowel wall contusion) with a reportedaccuracy of 75%–86%. CT was found to be less reli-able in determining the need for surgical interven-tion for mesenteric injuries, with an accuracy of54%–75% (40,41). Active bleeding and bowel wallthickening associated with mesenteric hematoma are

Figure 9. Active mesenteric hemorrhage.(a, b) Axial multi–detector row CT and (c) coro-nal multiplanar CT images show large mesen-teric hematoma (arrows) with active bleeding(black arrowheads) within center of hematoma.Hemoperitoneum (white arrowheads) is alsoseen. At surgery, mesenteric hemorrhage andhematoma were confirmed. Segment of is-chemic small bowel was also resected.

the most specific signs for surgically important me-senteric injury.

CT findings of a full-thickness bowel injury includeextraluminal gas (Fig 8) (17), intramural air, extralu-minal oral contrast material or intestinal content, anddiscontinuity of bowel wall. Focal bowel wall thicken-ing (Fig 8) (>4 mm) and isolated free peritoneal fluidare nonspecific CT findings of bowel injury (42,43).

CT findings of mesenteric injury include active ex-travasation of intravenous contrast material into themesentery (Fig 9), bowel wall thickening associatedwith mesenteric hematoma, focal mesenteric hema-toma (Fig 9), or mesenteric infiltration (Fig 8).

Radiologists should be familiar with abnormalitiesobserved in the bowel of patients with “hypoper-fusion complex” and fluid overresuscitation (44,45).These two entities can mask the diagnosis of bowelinjury (44,45). On CT images, small-bowel injuryusually appears as a segmental focal abnormality, incontrast to the diffuse abnormality involving the en-tire small bowel associated with both hypoperfusioncomplex and fluid overresuscitation. The bowel wallchanges seen following vigorous fluid overresuscita-tion may result in diffuse edema of the bowel wall,particularly the small bowel (45). Other CT findingsof intravenous volume expansion include periportallow attenuation, distention of the inferior vena cava,retroperitoneal fluid attenuation, and occasionallyascites, and these CT findings usually accompanybowel wall thickening.

The CT findings of shock bowel or diffuse small-bowel ischemia in hypotensive adult patients withblunt trauma include diffuse thickening of the small-bowel wall (range, 7–15 mm), fluid-filled dilated small-bowel loops, increased contrast enhancement of the

Shan

mu

gan

ath

an

110

small-bowel wall from slow perfusion and interstitialleakage of intravenous contrast material, and a flattenedinferior vena cava. The large bowel usually appears nor-mal. Usually, follow-up CT performed after adequatetreatment of the cause of hypoperfusion shows com-plete resolution of the small-bowel changes (44).

RENAL INJURY

CT provides information valuable to the diagnosis andstaging of blunt-force renal trauma in hemodynami-cally stable patients (46–48). An intravenous pyelo-gram can be obtained in the admitting area or the oper-ating room to depict both kidneys in hemodynamicallyunstable patients taken directly to surgery. Controversystill exists regarding the strength of the relationship be-tween hematuria and renal injury (46).

CT aids the accurate assessment of parenchymaldisruption, the integrity of the renal collecting sys-

tems, excretion of urine, the extent of perinephric he-matoma, and active hemorrhage. Management ofrenal injuries can be based on these CT findings andthe overall status of the patient.

CT of Minor Renal Injuries

Most renal injuries (75%–98%) are minor and aretreated without surgical intervention (49–51). Contu-

Figure 11. Cata-strophic renal in-jury. (a, b) Multi–detector row axialCT images showlarge perinephrichematoma(straight arrows)with fragmentationof renal paren-chyma (curved ar-rows). Perinephrichematoma dis-places peritonealorgans anteriorlyand to the left.(c) Coronal maxi-mum intensityprojection imageshows fragmentedrenal parenchyma(straight arrows)and renal arterybranch (curved ar-row) that supplieslower pole.

Figure 10.Renal lacera-tion and con-tusion.(a) Multi–detector rowaxial CT im-age obtainedat admissionshows renallaceration(straight ar-row) in midanterior por-tion of kidney;lacerationdoes not ex-tend into col-lecting sys-tem. Smallperinephrichematoma(curved whitearrow) andstranding(curved blackarrow) of fataround renalhilum are alsoseen. Follow-up (b) axialnonenhancedmulti–detectorrow CT imageand (c) coro-nal multi-planar refor-mation ob-tained 4hours afteradmissionshow collections of parenchymal extravasation (curved arrows) inareas of renal contusion that were not seen on CT images ob-tained at admission.

CT o

f Ab

do

min

al Traum

a: Part II

111

sions may appear as ill-defined low-attenuation areasor discretely limited regions of a striated nephro-graphic pattern from parenchymal extravasation ondelayed postcontrast CT images (Fig 10). Subcapsularhematomas are rare, assume a convex shape, com-press the renal parenchyma, and are limited by therenal capsule. Lacerations are seen on contrast-en-hanced CT images as linear low-attenuation areas(Fig 10a). In most cases, these injuries will resolvewithout intervention.

Segmental renal infarcts are relatively common fol-lowing blunt trauma and occur at the upper or lowerpole of the kidney. The infarcts appear as wedge-shaped, sharply demarcated low-attenuation areas.Isolated segmental renal infarcts do not warrant renalarteriography for confirmation or treatment.

CT of Major Renal Injuries

Major renal injuries include lacerations extendinginto the collecting system with urinary extravasation,large perinephric hematomas, renal lacerations dis-rupting more than 50% of the renal parenchyma, andsubcapsular hematomas causing delay in excretion(46). These lesions may or may not require surgical orangiographic intervention. Collecting system injurieswith persistent urine leakage may be treated with per-cutaneous nephrostomy or double-J ureteral catheter-ization. Angiographic embolization or surgical inter-vention may be required to treat expanding or largeperinephric or parenchymal hematomas.

CT of Catastrophic Renal Injuries

Catastrophic renal injuries warrant surgical orangiographic intervention. Evidence of these injuriesinclude CT findings of (a) fragmentation of renal pa-renchyma with associated large perinephric or para-renal hematoma (Fig 11), (b) major renal pedicle inju-ries (renal artery or vein) (Fig 12), (c) renal arterialbleeding or pseudoaneurysm, (d) renal pelvis disrup-tion, and (e) urinary leak into the peritoneal space.

Renal artery occlusion results from an intimal injuryat the junction of the proximal and middle one-thirdof the renal artery. Contrast-enhanced CT can usually

be used to confirm the diagnosis by demonstrating alack of renal enhancement (Fig 12) and diminishedkidney size. Peripheral subcapsular cortical enhance-ment may be seen from collateral vessels but is nottypically present in the acute setting. Angiography isnot required to establish the diagnosis and may delaydefinitive treatment.

CT findings of renal pelvis disruption include limitedor no parenchymal disruption and urinary contrast ex-travasation adjacent to the ureteropelvic junction andanterior pararenal space. The injury may be missed athelical CT unless delayed images are obtained with uri-nary contrast material within the renal pelvis. In thepresence of proximal urine extravasation, depiction ofthe distal ureter indicates a partial disruption.

References1. Fabian TC, Mangiante EC, White TJ, et al. A prospective study

of 91 patients undergoing computed tomography and peritoneallavage following blunt abdominal trauma. J Trauma 1986; 26:602–608.

2. Jones TK, Walsh JW, Maull KI. Diagnostic imaging in blunttrauma of the abdomen. Surg Gynecol Obstet 1983; 157:389–398.

3. Bain IM, Kirby RM, Tiwari P, et al. Survey of abdominal ultra-sound and diagnostic peritoneal lavage for suspected intra-abdominal injury following blunt trauma. Injury 1998; 29:65–71.

4. Peitzman AB, Makaroun MS, Slasky BS, et al. Prospectivestudy of computed tomography in initial management of bluntabdominal trauma. J Trauma 1986; 26:585–592.

5. Shanmuganathan K, Killeen KL. Imaging of abdominal trauma.In: Mirvis SE, Shanmuganathan K, eds. Imaging in trauma andcritical care. 2nd ed. Philadelphia, Pa: Saunders, 2003; 369–482.

6. Pachter HL, Guth AA, Hofstetter SR, Spencer FC. Changingpatterns in the management of splenic trauma: the impact onnonoperative management. Ann Surg 1998; 227:708–719.

7. Davis KA, Fabian TC, Corce MA, et al. Improved success innonoperative management of blunt splenic injury: embolizationof splenic artery pseudoaneurysm. J Trauma 1998; 44:1008–1015.

8. Shanmuganathan K, Mirvis SE, Boyd-Kranis R, Takada T,Scalea TM. Nonsurgical management of blunt splenic injury:use of CT criteria to select patients for splenic arteriographyand potential endovascular therapy. Radiology 2000; 217:75–82.

9. Meguid AA, Bair HA, Howells GA, Bendick PJ, Kerr HH, VillalbaMR. Prospective evaluation of criteria for the nonoperative man-agement of blunt splenic trauma. Am Surg 2003; 69:238–242;discussion 242–243.

Figure 12. Renal artery occlu-sion. (a) Axial multi–detector rowCT image and (b) coronal multi-planar reformation show no con-trast enhancement of right kid-ney (straight white arrows). Rightrenal artery is completely oc-cluded (black arrow) at junctionof proximal and middle segmentsof vessel. Hematoma (curvedwhite arrows) is seen around re-nal pedicle.

Shan

mu

gan

ath

an

112

10. Hartnett KL, Winchell RJ, Clark DE. Management of adult splenicinjury: a 20-year perspective. Am Surg 2003; 69:608–611.

11. Pimpl W, Dapunt O, Kaindl H, Thalhamer J. Incidence of septicand thromboembolic-related deaths after splenectomy inadults. Br J Surg 1989; 76:517–521.

12. Moore EE, Cogbill TH, Jurkovich GJ, et al. Organ injury scaling:spleen and liver (1994 revision). J Trauma 1995; 38:323–324.

13. Mirvis SE, Whitley NO, Vainwright JR, Gens DR. Blunt splenictrauma in adults: CT-based classification and correlation withprognosis and treatment. Radiology 1989; 171:33–39.

14. Starnes S, Klein P, Magagna L, et al. Computed tomographicgrading is useful in the selection of patients for nonoperativemanagement of blunt injury to the spleen. Am Surg 1998;64:743–748; discussion 748–749.

15. Federle MP, Courcoulas AP, Powell M, Ferris JV, Petizman AB.Blunt splenic injury in adults: clinical and CT criteria for man-agement, with emphasis on active extravasation. Radiology1998; 206:137–142.

16. Gavant ML, Schurr M, Flick PA, Croce MA, Fabian TC, GoldRE. Predicting clinical outcome of nonsurgical management ofblunt splenic injury: using CT to reveal abnormalities in thesplenic vasculature. AJR Am J Roentgenol 1997; 168:207–212.

17. Shanmuganathan K. Multi-detector row CT imaging of blunt ab-dominal trauma. Semin Ultrasound CT MR 2004; 25:180–204.

18. Shanmuganathan K, Mirvis SE, Sover ER. Value of contrast-enhanced CT in detecting active hemorrhage in patients withblunt abdominal or pelvic trauma. AJR Am J Roentgenol 1993;161:65–69.

19. Jeffrey RB Jr, Cardoza JD, Olcott EW. Detection of active ab-dominal arterial hemorrhage: value of dynamic contrast-en-hanced CT. AJR Am J Roentgenol 1991; 156:725–729.

20. Miller LA, Mirvis SE, Shanmuganathan K. CT diagnosis ofsplenic infarction in blunt trauma: imaging features, clinical sig-nificance and complications. Clin Radiol 2004; 59:342–348.

21. Matthes G, Stengel D, Seifert J, et al. Blunt liver injuries inpolytrauma: results from a cohort study with the regular use ofwhole-body helical computed tomography. World J Surg 2003;27:1124–1130.

22. Feliciano DV, Mattox KL, Jordan GL Jr, Burch JM, Bitondo CG,Cruse PA. Management of 1000 consecutive cases of hepatictrauma (1979-1984). Ann Surg 1986; 204:438–445.

23. Asensio JA, Demetriades D, Chahwan S, et al. Approach to themanagement of complex hepatic injuries. J Trauma 2000; 48:66–69.

24. Gao JM, Du DY, Zhao XJ. Liver trauma: experience in 348cases. World J Surg 2003; 27:703–708.

25. Asensio JA, Roldan G, Petrone P, et al. Operative manage-ment and outcomes in 103 AAST-OIS grades IV and V complexhepatic injuries: trauma surgeons still need to operate, butangioembolization helps. J Trauma 2003; 54:647–653; discus-sion 653–654.

26. Pachter HL, Feliciano DV. Complex hepatic injuries. Surg ClinNorth Am 1996; 76:763–782.

27. Mirvis SE, Whitley NO, Vainwright JR, Gen DR. Blunt hepatictrauma in adults: CT-based classification and correlation withprognosis and treatment. Radiology 1989; 171:27–32.

28. Becker CD, Gal I, Baer HU, Vock P. Blunt hepatic trauma inadults: correlation of CT injury grading with outcome. Radiology1996; 201:215–220.

29. Poletti PA, Mirvis SE, Shanmuganathan K, et al. CT criteria formanagement of blunt liver trauma: correlation with angiographicand surgical findings. Radiology 2000; 216:418–427.

30. Hagiwara A, Yukioka T, Ohta S, et al. Nonsurgical manage-ment of patients with blunt hepatic injury: efficacy of trans-catheter arterial embolization. AJR Am J Roentgenol 1997;169:1151–1156.

31. Savolaine ER, Grecos GP, Howard J, et al. Evolution of CTfindings in hepatic hematoma. J Comput Assist Tomogr 1985;9:1090–1096.

32. Denton JR, Moore EE, Coldwell DM. Multimodality treatment forgrade V hepatic injuries: perihepatic packing, arterial emboliza-tion, and venous stenting. J Trauma 1997; 42:964–967; discus-sion 967–968.

33. Buckman RF Jr, Miraliakbari R, Badellino MM. Juxtahepaticvenous injuries: a critical review of reported management strat-egies. J Trauma 2000; 48:978–984.

34. Yokota J, Sugimoto T. Clinical significance of periportal track-ing on computed tomographic scan in patients with blunt livertrauma. Am J Surg 1994; 168:247–250.

35. Shanmuganathan K, Mirvis SE, Amerosa M. Periportal lowdensity on CT in patients with blunt trauma: association withelevated venous pressure. AJR Am J Roentgenol 1993; 160:279–283.

36. Williams MD, Watts D, Fakhry S. Colon injury after blunt ab-dominal trauma: results of the EAST Multi-Institutional HollowViscus Injury Study. J Trauma 2003; 55:906–912.

37. Rizzo MJ, Federle MP, Griffiths BG. Bowel and mesenteric in-jury following blunt abdominal trauma: evaluation with CT. Ra-diology 1989; 173:143–148.

38. Donohue J, Crass R, Trunkey D. Management of duodenal andsmall intestinal injury. World J Surg 1985; 9:904–913.

39. Malhotra AK, Fabian TC, Katsis SB, et al. Blunt bowel and mes-enteric injuries: the role of screening computed tomography. JTrauma 2000; 48:991–998.

40. Killeen KL, Shanmuganathan K, Poletti PA, Cooper C, MirvisSE. Helical computed tomography of bowel and mesenteric in-juries. J Trauma 2001; 51:26–36.

41. Dowe MF, Shanmuganathan K, Mirvis SE, et al. CT findings ofmesenteric injury after blunt trauma: implications for surgical in-tervention. AJR Am J Roentgenol 1997; 168:425–428.

42. Hanks PW, Brody JM. Blunt injury to mesentery and smallbowel: CT evaluation. Radiol Clin North Am 2003; 41:1171–1182.

43. Hawkins AE, Mirvis SE. Evaluation of bowel and mesenteric in-jury: role of multidetector CT. Abdom Imaging 2003; 28:505–514.

44. Mirvis SE, Shanmuganathan K, Erb R. Diffuse small-bowel is-chemia in hypotensive adults after blunt trauma (shock bowel):CT findings and clinical significance. AJR Am J Roentgenol1994; 163:1375–1379.

45. Chamrova Z, Shanmuganathan K, Mirvis SE, et al. Retroperito-neal fluid resulting from rapid intravascular resuscitation intrauma: CT mimic of retroperitoneal injury. Emerg Radiol 1994;1:85–88.

46. Mirvis SE. Injuries to the urinary system and retroperitoneum. In:Mirvis SE, Shanmuganathan K, eds. Imaging in trauma and criti-cal care. 2nd ed. Philadelphia, Pa: Saunders, 2003; 483–518.

47. Stein JP, Kaji DM, Eastham J, Freeman JA, Esrig D, Hardy BE.Blunt renal trauma in the pediatric population: indications for ra-diographic evaluation. Urology 1994; 44:406–410.

48. Smith JK, Kenney PJ. Imaging of renal trauma. Radiol ClinNorth Am 2003; 41:1019–1035.

49. Toutouzas KG, Karaiskakis M, Kaminski A, et al. Nonoperativemanagement of blunt renal trauma: a prospective study. AmSurg 2002; 68:1097–1103.

50. Matthews LA, Smith EM, Spirnak JP. Nonoperative treatment ofmajor renal lacerations with urinary extravasation. J Urol 1997;157:2056–2058.

51. Robert M, Drianno, Muir G, et al. Management of major bluntrenal lacerations: surgical or nonoperative approach? Eur Urol1996; 30:335–339.

113

Imaging thePediatric Patient with

Acute Abdominal Disease1

The most frequent nontraumatic conditions resulting in an acute abdomen in childrenare midgut malrotation, intestinal intussusception, and acute appendicitis. The imagingevaluation of these conditions strongly affects diagnosis and management. Therefore, se-lection of an appropriate imaging strategy is essential to ensure prompt treatment. Thischapter focuses on the important imaging features of these three conditions.

MIDGUT MALROTATION

Midgut malrotation is the most important cause of upper intestinal obstruction in new-borns. The malfixated bowel is associated with a narrow mesentery that is prone to twist-ing, which results in occlusion of mesenteric vessels and secondary bowel necrosis. Thetwisting is labeled a midgut volvulus. In addition, aberrant peritoneal bands are frequentlypresent. These are called Ladd bands and can lead to duodenal obstruction.

The clinical manifestation of midgut malrotation is usually characterized by vomit-ing, which is often bilious. Note that the majority of patients with bilious vomiting donot have malrotation (1). Symptoms may occur at any age, but most patients presentin the 1st month of life.

Evaluation of infants and children who are suspected of having midgut malrotationtypically begins with abdominal radiography. However, the radiographic findings as-sociated with malrotation are nonspecific. The abdominal radiograph may demon-strate a proximal obstruction caused by Ladd bands or a distal small-bowel obstruc-tion caused by a midgut volvulus (2). However, the abdominal radiograph may shownormal findings. Therefore, radiography should not be used to exclude the diagnosisof malrotation.

The examination of choice for the diagnosis of malrotation is the upper gastrointes-tinal examination. Midgut malrotation with malfixation of the intestines is inferredfrom malposition of the duodenojejunal junction. The normal location of the duode-nojejunal junction is to the left of the spine (to the left of the left pedicles) and at thelevel of the duodenal bulb (3). Patients with malrotation have an abnormally posi-tioned duodenojejunal junction located to the right or inferior to the normal position(Fig 1). In a child without obstruction, the abnormally located duodenojejunal junc-tion may be the only finding. One diagnostic pitfall to avoid is the fact that an over-distended stomach may inferiorly displace the normal duodenojejunal junction

Carlos J. Sivit, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 113–117.

1 From the Departments of Radiology and Pediatrics, Rainbow Babies and Children’s Hospital of the University Hospitalsof Cleveland and Case Western Reserve University School of Medicine, 11100 Euclid Ave, Cleveland, OH 44106-5056(e-mail: sivit@uhrad.com).

Sivi

t

114

below its normal location. In such circumstances, re-peating the upper gastrointestinal examination by ad-ministering a small amount of contrast material (10–15 mL) through a nasoenteric tube can help to confirmthe diagnosis.

The diagnosis of midgut malrotation can also bemade at ultrasonography (US) and computed tomogra-phy (CT). Most patients with malrotation have inver-sion of the normal relationship between the mesentericvessels (4). Thus, the superior mesenteric vein lies tothe left of the superior mesenteric artery (Fig 2). How-ever, this finding is neither as sensitive nor as specific asthe identification of the duodenojejunal junction onimages from the upper gastrointestinal series (4).Therefore, the diagnosis should always be confirmedwith an upper gastrointestinal examination.

INTUSSUSCEPTION

Intestinal intussusception occurs when a segment ofbowel, the intussusceptum, telescopes into a more dis-tal segment, the intussuscipiens. Intussusception oc-curs more frequently in boys than in girls. It is rare ininfants younger than 3 months and in children olderthan 3 years of age. The peak incidence is between 5and 9 months of age. The most common type of intus-susception is ileocolic, followed by ileoileocolic. Ap-proximately 95% of all pediatric intussusceptions haveno pathologic lead point and result from hypertrophyof lymphoid tissues, typically following a recent viralinfection. In 5% of the patients, recognizable causes forthe intussusception are found, including Meckel diver-ticulum, intestinal polyp, enteric duplication, intramu-ral hematoma, and lymphoma.

The clinical manifestation is usually characterizedby intermittent colicky abdominal pain. Vomiting is

also commonly noted. Lethargy and somnolence maydevelop later. Stool containing blood and mucus, oth-erwise described as currant jelly stool, may be noted inapproximately two-thirds of the patients. Abdominaldistention and tenderness may develop if the disorderis complicated by bowel obstruction.

Evaluation of children suspected of having intus-susception has traditionally begun with abdominalradiography. The role of abdominal radiography isto serve as a screening examination for the detectionof intussusception and to assess for possible compli-cations related to the condition, such as bowel obstruc-tion or perforation. Radiographic findings of intussus-ception include (a) a soft-tissue mass or (b) no depic-tion of an air-filled right colon. A distal small-bowelobstruction and extraluminal air may also be noted,caused by complications of the condition. Abdominalradiographs may be normal in nearly one-half of allchildren with intussusception (5). Therefore, the diag-nosis cannot be excluded on the basis of the radio-graphic findings.

US is being used with increasing frequency for the di-agnosis of intussusception. A technique of graded com-pression is used, with pressure slowly applied with thetransducer during the examination. The results of nu-merous studies have shown that US has a high sensitiv-ity and specificity for this diagnosis. Therefore, it maybe confidently used as the primary screening examina-tion (6–8). At our hospital, US has replaced radiogra-phy as the primary screening modality for children sus-pected of having intussusception. The US appearance ofan intussusception is an outer hypoechoic ring with ahyperechoic center or multiple concentric rings. On thetransverse view, the appearance has been likened to adoughnut (Fig 3), whereas on longitudinal section, theappearance has been likened to a “pseudokidney.”

Figure 2. Midgut malrotation. Transverse sonogram throughthe middle of the abdomen of a child with malrotation shows in-version of the mesenteric vessels, with the superior mesentericvein (V) to the left of the superior mesenteric artery (A).

Figure 1. Midgut malrotation. Spot radiograph from an uppergastrointestinal examination demonstrates the duodenojejunaljunction in an abnormal location to the right of the spine and be-low the duodenal bulb. This finding is diagnostic for malrotation.

Pediatric A

cute A

bd

om

inal D

isease

115

The therapeutic study of choice for intussusception isa contrast enema study. Rates of reduction of intussus-ception range from 60% to 90% (9,10). Perforation isreported in approximately 1% of the attempts at reduc-tion. The only absolute contraindication to attemptingto reduce an intussusception with imaging guidance isperforation. Successful reduction of intussusception ismarked by the free flow of contrast material into thedistal ileum, associated with disappearance of the soft-tissue mass.

Controversy exists about the optimal choice of thecontrast material to use for the treatment of intussuscep-tion: Air (Fig 4), water-soluble contrast material (Fig 5),or barium may be used. Proposed advantages of intus-susception reduction with air include decreased cost, de-creased fluoroscopy time and radiation dose, ability tomonitor the intraluminal pressure generated, and lessfecal spillage if perforation occurs. Proposed advantagesof reduction with liquid contrast material include im-proved depiction of the intussusception and of contrastmaterial refluxing into the small intestine and earlierdepiction of perforation. If water-soluble liquid contrastmaterial is used, a dilute meglumine–sodium diatri-zoate mixture is preferred. A 1:3 or 1:4 dilution withwater results in a relatively iso-osmolar concentration.

ACUTE APPENDICITIS

Acute appendicitis is the most common condition re-quiring emergency abdominal surgery in the pediatricpopulation (11). The condition typically develops inolder children and young adults. The incidence in the

Figure 4. Intussusception. Spot radiograph from an air-contrastenema study shows a focal mass representing an intussuscep-tion in the left upper quadrant.

Figure 3. Intussusception. Transverse sonogram through themiddle of the abdomen demonstrates a focal rounded mass rep-resenting an intussusception.

Figure 5. Intussusception. Spot radiograph from a water-soluble contrast enema study demonstrates an intussusceptionin the ascending colon.

Sivi

t

116

pediatric population is highest in male patients be-tween the ages of 10 and 14 years, while in female pa-tients, the incidence is highest between the ages of 15and 19 years (12).

Acute appendicitis presents a challenging problemto caregivers because it must be differentiated from avariety of other conditions that result in acute ab-dominal pain in childhood. The classic constellationof periumbilical pain migrating to the right lowerquadrant, nausea, vomiting, and fever is present inless than one-third of the patients (13). The diagnosisis even more challenging in younger children becausethey are not able to clearly describe their symptoms.Between one-third and one-half of the children withappendicitis who undergo surgery have an uncertainpreoperative diagnosis (13). Often, many of thesechildren are initially admitted for observation prior toundergoing surgery.

Abdominal radiography has been shown to be arelatively insensitive and nonspecific means to evalu-ate appendicitis (14). In nonperforated appendicitis,the abdominal radiograph is usually normal or dem-onstrates nonspecific findings such as diffuse air-fluidlevels or mild bowel dilatation. Therefore, the routineuse of abdominal radiography has little value unlessbowel obstruction or perforation is suspected.

There has been a great deal of variability in the useof cross-sectional imaging for the diagnosis of acuteappendicitis in children. In the period since its intro-duction by Puylaert (15) in 1986 up until the mid-1990s, graded-compression US has been the principalimaging technique for evaluating children suspectedof having appendicitis (13,16). Operator skill is animportant factor in the diagnostic accuracy of US, asevidenced by the great variability in the reported diag-nostic sensitivity and specificity of the examination.

The reported sensitivity of US in children has rangedfrom 44% to 100%, and the specificity has rangedfrom 47% to 95%.

The inflamed appendix appears at US as a fluid-filled noncompressible blind-ending structure mea-suring greater than 6 mm in maximal diameter. Otherfindings of appendicitis include (a) an appendicolith,which appears as a hyperechoic focus with acousticshadowing, (b) pericecal or periappendiceal fluid, and(c) increased periappendiceal echogenicity, represent-ing fat infiltration. The characteristic US findings asso-ciated with perforated appendicitis include (a) focalperiappendiceal or pelvic fluid collections or (b) acomplex mass, representing intraperitoneal abscess.

CT has become the modality of choice in the evalu-ation of children suspected of having appendicitis. CThas been shown to be a highly sensitive and specificexamination for the diagnosis of acute appendicitis inchildren. CT is less dependent on the operator than isUS, and thus, higher diagnostic accuracies have beenachieved. The reported sensitivity of CT for the diag-nosis of acute appendicitis in children has rangedfrom 87% to 100%, and the specificity has rangedfrom 89% to 98% (17–20). CT is also more usefulthan US for evaluating complications of acute appen-dicitis, such as phlegmon and abscess formation.

A variety of CT techniques have been used in the per-formance of appendiceal CT. These techniques include(a) full abdominopelvic scanning after intravenous andoral administration of contrast material, (b) imaginglimited to the lower portion of the abdomen and pelviswithout any contrast material, (c) imaging limited to

Figure 7. Acute appendicitis. Contrast-enhanced CT scanthrough the lower part of the abdomen in a child with appendicitisshows a distended appendix with an appendicolith.

Figure 6. Acute appendicitis. Contrast material–enhancedCT scan through the lower part of the abdomen of a child withappendicitis shows an enlarged appendix. Note the appen-diceal wall enhancement and surrounding stranding of peri-appendiceal fat.

Pediatric A

cute A

bd

om

inal D

isease

117

the lower portion of the abdomen and pelvis with theuse of orally and rectally administered contrast materi-al, and (d) imaging of the lower portion of the abdo-men and pelvis with the use of only rectally adminis-tered contrast material. The goals in using rectally ad-ministered contrast material are to (a) distend thesigmoid colon and cecum, (b) delineate the cecal walland identify wall thickening, and (c) opacify the ap-pendix if it is not obstructed. A 3% diatrizoate meg-lumine solution is administered intracolonicallythrough a rectal catheter. The administered volumeof fluid ranges from 500 mL in small children to1000 mL in adolescents. Intravenous administrationof contrast material can also aid in the diagnosis ofappendicitis by permitting the identification of theinflamed appendix and allowing differentiation ofthe appendix from the adjacent iliac vessels.

CT signs of acute appendicitis include a distendedappendix greater than 7 mm in maximal diameter(Fig 6), appendiceal wall thickening and enhance-ment (Fig 6), an appendicolith (Fig 7), circumferen-tial or focal apical cecal thickening, pericecal fatstranding, adjacent bowel wall thickening, focal orfree peritoneal fluid, mesenteric lymphadenopathy,intraperitoneal phlegmon, or abscess (Fig 8). Theidentification of cecal apical thickening is particularlyuseful in allowing a confident diagnosis of acute ap-pendicitis if there is difficulty in identifying an en-larged appendix.

References1. Lilien L, Srinivasan G, Pyatt S, et al. Green vomiting in the

first 72 hours in normal infants. Am J Dis Child 1986; 140:662–664.

2. Berdon WE, Baker DH, Bull S, Santulli TV. Midgut malrota-tion and volvulus: which films are most helpful? Radiology1970; 96:375–384.

3. Long FR, Kramer SS, Markowitz RI, Taylor GE, LiacourasCA. Intestinal malrotation in children: tutorial on radiographicdiagnosis in difficult cases. Radiology 1996; 198:775–780.

4. Zerin JM, DiPietro MA. Superior mesenteric vascular anato-my at US in patients with surgically proved malrotation ofthe midgut. Radiology 1992; 183:693–694.

5. Sargent MA, Babyn P, Alton DJ. Plain abdominal radiogra-phy in suspected intussusception: a reassessment. PediatrRadiol 1994; 24:17–20.

6. Lim HK, Bae SH, Lee KH, Seo GS, Yoon GS. Assessmentof reducibility of ileocolic intussusception in children: useful-ness of color Doppler sonography. Radiology 1994; 191:781–785.

7. Verschelden P, Filiatrault D, Garel L, et al. Intussusceptionin children: reliability of US in diagnosis—a prospectivestudy. Radiology 1992; 184:741–744.

8. Weinberger E, Winters WD. Intussusception in children: therole of sonography. Radiology 1992; 184:601–602.

9. Eklof OA, Johanson L, Lohr G. Childhood intussusception:hydrostatic reducibility and incidence of leading points indifferent age groups. Pediatr Radiol 1980; 10:83–86.

10. Gu L, Alton DJ, Daneman A, et al. John Caffey Award: in-tussusception reduction in children by rectal insufflation ofair. AJR Am J Roentgenol 1988; 150:1345–1348.

11. Janik JS, Firor HV. Pediatric appendicitis: a 20-year studyof 1,640 children at Cook County (Illinois) Hospital. ArchSurg 1979; 114:717–719.

12. Bell MJ, Bower RJ, Ternberg JL. Appendectomy in child-hood: analysis of 105 negative explorations. Am J Surg1982; 144:335–337.

13. Sivit CJ, Newman KD, Boenning DA, et al. Appendicitis:usefulness of US in diagnosis in a pediatric population. Ra-diology 1992; 185:549–552.

14. Campbell JP, Gunn AA. Plain abdominal radiographs andacute abdominal pain. Br J Surg 1988; 75:554–556.

15. Puylaert JB. Acute appendicitis: US evaluation usinggraded compression. Radiology 1986; 158:355–360.

16. Vignault F, Filiatrault D, Brandt ML, Garel L, Grignon A,Ouimet A. Acute appendicitis in children: evaluation withUS. Radiology 1990; 176:501–504.

17. Garcia-Pena BM, Mandl KD, Kraus SJ, et al. Ultrasonogra-phy and limited computed tomography in the diagnosis andmanagement of appendicitis in children. JAMA 1999; 282:1041–1046.

18. Lowe LH, Penney MW, Stein SM, et al. Unenhanced limitedCT of the abdomen in the diagnosis of appendicitis in chil-dren: comparison with sonography. AJR Am J Roentgenol2001; 176:31–35.

19. Sivit CJ, Applegate KE, Stallion A, et al. Imaging evaluationof suspected appendicitis in a pediatric population: effec-tiveness of sonography versus CT. AJR Am J Roentgenol2000; 175:977–980.

20. Sivit CJ, Siegel MJ, Applegate KE, Newman KD. When ap-pendicitis is suspected in children. RadioGraphics 2001; 21:247–262.

Figure 8. Acute appendicitis. Contrast-enhanced CT scanthrough the lower part of the abdomen of a child with a rupturedappendix demonstrates multiple intraloop fluid collections.

NO

TES

119

The Role of CT in AcuteAbdominal Disease: Pitfalls

and Their Lessons1

Learn from the mistakes of others. You can’t live long enough to make them all yourself.—Eleanor Roosevelt (1)

Since the introduction of the first commercial computed tomographic (CT) scanner in1973, the ability of single-section CT to help in the diagnosis of nontraumatic acutegastrointestinal disease, such as the different causes of intestinal obstruction, appendi-citis, diverticulitis, and mesenteric ischemia, has emerged. The introduction of helicalCT in 1989, the twin detector in 1991, and, subsequently, multi–detector row helicalCT technology in 1998 began to further change the way radiologists looked at the in-testinal tract. In addition to looking at solid organs, even more attention is now givento the intestines, mesentery, and visceral blood supply. The ability of CT to help diag-nose acute nontraumatic and trauma-related intestinal disorders and how CT haschanged clinical management are no longer in question (2).

CT is now the dominant method in the investigation of the patient with acute ab-dominal pain in the United States. The radiology literature is replete with reports onthe relevance, the high accuracy, and the role of CT in the diagnosis of causes of theacute abdomen and its management. Little, if any, emphasis, however, is placed onthe limitations and pitfalls of CT. This chapter reviews our observations on these is-sues and addresses pitfalls and limitations relevant to the diagnosis of small-boweldiseases, appendicitis, colonic diverticulitis, and intestinal ischemia. Recommenda-tions on how to decrease or avoid these shortcomings are discussed.

CLASSIFICATION OF ERRORS

The classification of radiologic errors is considered in the framework suggested inprior reports (3,4). This classification includes the following:

1. Perceptive error.—The abnormality is obvious in retrospect, and purely technicaldeficiencies of the examination were not thought to contribute to the error.

2. Interpretive error.—The lesion was observed at the time of the study, but the diag-nosis was not considered in the differential diagnosis, or the observation was thoughtto be insignificant or a normal variant.

Dean D. T. Maglinte, MD, James T. Rhea, MD,and M. Stephen Ledbetter, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 119–131.

1From the Departments of Radiology, Indiana University School of Medicine, UH0279, Indiana University Hospital, 550N University Blvd, Indianapolis, IN 46202-5253 (D.D.T.M.); Massachusetts General Hospital, Harvard Medical School,Boston, Mass (J.T.R.); and Brigham and Women’s Hospital, Harvard Medical School, Boston, Mass (M.S.L.) (e-mail:dmaglint@iupui.edu).

Mag

linte

et a

l

120

3. Combined perceptive and technical error.—An ab-normality was seen on review, and technical inad-equacy was thought to be contributory.

4. Technical limitation.—Lesions not visible in retro-spect were regarded as a technical limitation or“missed” lesion beyond the capability of the methodto demonstrate.

INTERRELATIONSHIPS OF ABDOMINALVISCERAL ORGANS AND THE INFLUENCEOF PERITONEAL FLUID DYNAMICS

Errors in the diagnosis of abdominal abnormalities atCT examination can be ascribed to inherent limitationsof the method, failure to optimize technique, failure toappreciate normal anatomic structures, and congenitalanatomic and postoperative alterations resulting in in-terpretive and perceptive errors. An understanding ofperitoneal fluid dynamics and anatomic structures ofthe small bowel, the colon and its mesentery, bloodsupply, and peritoneal reflections, and the variations inposition and size of the appendix are important in un-derstanding the limitations and pitfalls in the imaging

of the patient who presents with acute abdominal pain.Awareness of these interrelationships prevents misdiag-nosis and misleading false-negative and false-positiveinterpretations.

A small amount of serous fluid (50–100 mL) is usu-ally found in the peritoneum. This fluid circulateswithin the peritoneal cavity cephalad to caudad andback cephalad. Flow is governed by gravity (down-ward) and respiration (upward). During inspiration,the negative intrathoracic pressure is released, and fluidtravels downward by gravity (5). Negative pressure un-der the diaphragm during expiration causes peritonealfluid to move superiorly. Bowel peristalsis and the peri-toneal reflections and mesentery direct the fluid paths.These peritoneal reflections and recesses provide water-sheds and a drainage basin for the spread and localiza-tion of fluid (6). Thus, inflammatory processes involv-ing any segment of the alimentary tube may localizeelsewhere in the peritoneal cavity and be mistaken for aprimary abnormality unless these interrelationshipsand fluid dynamics are understood. Unrecognized rota-tional anomalies of the small bowel and colon andvariations in the size and position of the appendix, as

Figure 1. Perceptive and interpretive errorsecondary to inherent limitation of method ofexamination. (a, b) Axial CT scans of lowerportion of abdomen of a 38-year-old patientwith chronic renal failure who presented toemergency department with severe abdomi-nal pain, nausea, and vomiting. Three previ-ous abdominal CT scans (not shown) ob-tained in past 2 years because of similarsymptoms were unremarkable, as were re-sults of small-bowel follow-through examina-tion performed electively. Note oral contrastmaterial in small bowel and colon. Oblitera-tion of fat plane in right paramedian regionposterior to rectus muscle between anteriorsmall-bowel loops and anterior parietal peri-toneum is difficult to appreciate, in absenceof a transition point suggestive of small-bowel obstruction. (c, d) CT enteroclysis scans at same level as in a and b, obtained electively after conventional CT, show transi-tion point (arrow) and collapsed loops distally. Cause of obstruction is clearly defined. Right lower abdominal anterior enteroparietalperitoneal adhesions were evidenced by fixation noted on fluoroscopic phase of CT enteroclysis examination (not shown) and by lossof fat plane between anterior wall of flattened small bowel and adjacent parietal peritoneal lining (arrow). Compare with convex anteriormargin of small bowel on left side. Also note enteroenteric adhesions posterior to peritoneal adhesion in d. (e) Coronal CT reformationobtained at same level shows precise level of transition point in anterior portion of abdomen (information of value to laparoscopic sur-geons). Laparoscopic adhesiolysis confirmed findings after CT enteroclysis.

CT in

Acu

te Ab

do

min

al Disease

121

well as nonspecific imaging findings, contribute toboth perceptive and interpretive errors.

PITFALLS IN DIAGNOSIS

Small-Bowel Diseases

Symptoms of small-bowel disease can be mimickedby diseases of other organs. Because of the dimen-sions and anatomic location of the small bowel, dis-eases of adjacent organs may mimic intrinsic small-bowel disease (7).

Lack of understanding of the normal physiologicfunction of the mesenteric small intestine can lead tomisinterpretations. Filling and distention of the smallbowel are dependent on gastric emptying and thesmooth muscle of the intestinal wall to move intestinalcontents along the length of the bowel. The volume in-gested, the rate of gastric emptying, and diminishedsmall-bowel peristalsis may result in underfilling of thesmall intestine or stasis of orally ingested fluid. This isan acknowledged inherent limitation of examinationsperformed without enteral volume challenge, such asconventional radiography, the oral small-bowel exami-nation, or abdominal CT, in the diagnosis of lowergrades of mechanical small-bowel obstruction (8,9).

Intestinal obstruction remains a difficult entity todiagnose accurately and treat (8). Radiologists as-sume a critical role in the clinical decision makingfor patients with suspected or known small-bowelobstruction because imaging can provide answers tospecific questions that have a major effect on clinicalmanagement (10). Pitfalls are primarily combinedperceptive and interpretive errors secondary to thisinherent limitation of the methods of examination(Fig 1). Because of the serpentine course of the intes-tine in a limited space, lesions may not be apparentwhen viewed in a single plane. Multiplanar reformat-ting should be used. Simple mechanical obstructioncannot be reliably differentiated clinically from stran-gulated obstruction on the basis of the findings atphysical examination, laboratory results, or abdominalconventional radiographic findings (11). A prompt andprecise CT diagnosis allows triage of these patients intoa surgical or nonsurgical management category and de-creases the number of subsequent diagnostic examina-tions, the morbidity, and the cost of patient care.

When small-bowel stasis occurs, admixture defectsresult, especially when inadequate amounts of radio-paque oral contrast material (water-soluble contrastmaterial or dilute barium) are ingested or when con-trast material is vomited by a nauseated patient. Ab-normal images are produced (Fig 2). Because CT doesnot allow real-time assessment (static imaging), nor-mal physiologic features or enteric debris may simu-late pathologic findings. In addition to pseudomasses,admixture defects from stasis can also result inpseudo–bowel wall thickening. These defects result ininterpretive errors and may result in unnecessarywork-up for inflammatory bowel diseases (Fig 3).Medication-related hypoperistalsis or functional mo-tility abnormalities from prior surgery may simulateacute small-bowel obstruction. To avoid these errors,appropriate clinical correlation of the CT findings tothe medication history or prior surgery is necessary.

Congenital anatomic and postoperative alterationsmay result in perceptive and interpretive errors. Anunderstanding of midgut rotational anomalies andknowledge of the surgical history are important inavoiding these pitfalls (Fig 4).

Not infrequently, perforated appendicitis is misdi-agnosed as inflammatory bowel disease, ileus, oracute small-bowel obstruction because of the effectsof peritoneal irritation (Fig 5). Therefore, appendicitisshould be taken into consideration when acute find-ings of enteritis or acute small-bowel obstruction areseen in patients without a clinical prodrome of Crohndisease. Extraappendiceal abscess in the subacutephase may simulate an intrinsic small-bowel mass.The combination of a nonspecific clinical manifesta-tion of many small-bowel diseases, a misleading clini-cal history, and suboptimal bowel opacification re-sults in these interpretive errors (Fig 6).

Figure 2. Inter-pretive error sec-ondary to admix-ture defect simu-lating small-bowelmass. (a) Axial CTscan of abdomenof middle-agedpatient with nau-sea, vomiting, andabdominal painshows incom-pletely opacifiedlumen with pre-sumed mass andnarrowed bowellumen (arrow).Also note unopaci-fied proximalloops. (b) Radio-graph obtained atenteroclysis showsno small-bowelmass. The patientwas receiving nar-cotics for paincaused by chronicpancreatitis.

Mag

linte

et a

l

122

Figure 3. Interpretive error secondaryto abnormal peristalsis and admixture de-fects. (a, b) Axial CT scans of mid andlower portions of abdomen of middle-agedwoman with previous distal small-bowel re-section and partial colectomy for Crohn dis-ease who presented with severe abdominalpain. Findings of thickened walls of distalsmall-bowel loops and colon were inter-preted as consistent with recurrent Crohndisease. (c, d) Selected radiographs of abarium enteroclysis performed followingCT. No evidence of recurrence was seen.Note fluid and debris in dilated small-bowelloop (arrow in c) diluting contrast material.(d) Double-contrast barium enteroclysis im-age shows normal-sized folds. Results ofcolonoscopy and ileoscopy were also nor-mal. The patient was also addicted to nar-cotics.

Figure 5. Interpretive error secondary toperitoneal irritation and anatomic interrela-tionships in patient with abdominal pain andconstipation. Axial CT scans obtained withintravenous contrast material show mildthickening of distal small-bowel loops, mu-cosal hyperemia, and perienteric and mes-enteric stranding. Patient vomited oral con-trast material. Diagnosis was inflammatorybowel disease. Because of peritoneal signsand symptoms, surgery was performed,which revealed a perforated appendix withperitonitis.

Figure 4. Perceptive and interpretive er-ror secondary to congenital anatomic ab-normality (ie, unusual course of appendix).(a) Axial CT scan obtained without oral con-trast material in patient with severe abdomi-nal pain shows mesenteric stranding andarea of soft-tissue attenuation in midportionof abdomen. Presumed normal appendix isseen to the right of area of soft-tissue attenu-ation. Appendicolith within area of soft-tissueattenuation was not appreciated. Preliminarydiagnosis was inflammatory bowel disease.(b) CT scan obtained after opacification ofsmall bowel with oral contrast material showsappendicolith surrounded by soft-tissue attenuation and mesenteric stranding away from cecum. Surgery revealed abscess with fecalithat appendiceal tip, located in midportion of abdomen. (Images courtesy of Bernard Birnbaum, MD, New York, NY.)

CT in

Acu

te Ab

do

min

al Disease

123

Appendicitis

CT has had an important effect on the manage-ment of patients suspected of having appendicitis(12,13). Perceptive and interpretive errors may re-sult from secondary involvement of small bowel byperforative appendicitis. Such secondary involve-ment mimics primary small-bowel disorders be-cause of peritoneal dynamics and anatomic interre-lationships. Among the specific problems encoun-tered in the diagnosis of appendicitis is a failure toopacify the cecum and appendix. These commontechnical pitfalls and limitations result in no depic-tion of the normal appendix, which increases false-negative or false-positive diagnoses (Figs 5, 6) andlowers the diagnostic certainty of a normal exami-nation. Such failure also results in the inability todepict the findings that are 100% specific for appen-dicitis: focal cecal apical thickening, the “arrow-head” sign, and the cecal bar adjacent to an appen-dicolith (14–17). Inadequate familiarity with thefindings of appendicitis results in perceptive error.Nonspecific features may be considered conclusiveand specific findings and can be misapplied if thececal apex is not identified. A learning curve existsin recognizing either the abnormal or the normalappendix (15,16).

Lack of intraabdominal fat, especially in children,the elderly, and cachectic patients, increases the diffi-culty of finding the normal or abnormal appendix.Sonography, when expertise is available, is an impor-

tant examination in this subset of patients, particularlyin children. The CT technique should always includethe intravenous administration of contrast material inthese patients because of the resulting enhancement ofthe abnormal appendiceal wall and the demonstrationof skip areas in the enhancement when the appendix isfocally necrotic (14–17). A failure to monitor the ex-amination and administer additional contrast materialor to obtain different projections if needed to helpclarify initially indeterminate findings is a pitfall thatcan be avoided. Volume averaging associated with CTsection thickness greater than 5 mm can limit the abil-ity to distinguish the appendix from adjacent struc-tures. Failure to use appropriate CT window settings,especially if free fluid is present, may obscure the ap-pendix and result in interpretive errors. Interpretive er-rors also result when the diagnostic threshold is toolow for the CT diagnosis of acute appendicitis (Fig 7).

Variation in the position of the cecum or appen-dix may cause difficulty in interpretation (Fig 4).The transverse cecum, cecal bascule, or malpositionof the colon makes it imperative to identify the posi-tion of the cecum and terminal ileum before at-tempting to identify the appendix. Other interpretiveproblems include misidentification of an opacifiedvessel for the normal appendix or misidentificationof a loop of unopacified small bowel for the abnor-mal appendix. Tip appendicitis, stump appendicitis,and secondary appendicitis may all present interpre-tive problems. As discussed in the section on the

Figure 6. Interpretive error influenced bymisleading clinical history, nonspecific clini-cal manifestation, paucity of abdominal fat,and effects of peritoneal irritation on distalsmall bowel from pus from perforated ap-pendix. (a, b) Initial axial CT scans of teen-age male patient with severe abdominal painand family history of Crohn disease showextensive mesenteric stranding without evi-dence of pericecal inflammatory changes(not shown). There is paucity of mesentericfat and no contrast material filling of pelvicsmall-bowel loops. (c, d) Delayed axial CTscans show increased mesenteric attenua-tion and poorly opacified pelvic loops withmarkedly thickened loops of pelvic ileum.Oral contrast material was vomited. Exami-nation was interpreted as worrisome for ac-tive Crohn disease, and enteroclysis wasrecommended. Patient was referred to gas-troenterologist, who instead performedcolonoscopy, which yielded unremarkablefindings. Worsening of symptoms and a fol-low-up CT scan (not shown) that revealedpelvic abscesses prompted surgery. Surgeryconfirmed pelvic abscesses from a perfo-rated appendix.

Mag

linte

et a

l

124

small bowel, findings secondary to appendicitis areoften attributed to primary small-bowel diseases(Fig 8). There are a variety of alternative diagnoses(or mimics), which should be sought because of thenonspecific clinical findings in appendicitis. Inde-terminate CT examinations should be resolved byopacification of the cecum and appendix in theemergency department. A misleading clinical historyalso results in an interpretive pitfall (Fig 9).

Diverticulitis

Diverticulitis eventually complicates approximately20% of the cases of colonic diverticulosis. Diverticularinflammation with perforation results in an intramu-ral or a localized pericolic abscess. Complications in-clude bowel obstruction, bleeding, peritonitis, and si-nus tract and fistula formation. Interpretive errors oc-cur secondary to a nonspecific clinical manifestation,subtle findings early in the manifestation, or limita-

Figure 7. Interpretive error caused by setting threshold too low in CT diagnosis of acute appendicitis. Axial CT scans at level ofcecum of patient with acute abdominal pain show 8-mm-wide appendix without periappendiceal stranding but with adjacent enlargednodes. CT findings were interpreted as indeterminate. Surgery and pathologic evaluation of specimen showed acute appendicitis.

Figure 8. Interpretive pitfall: misinterpreta-tion of secondary finding caused by appendi-citis as the principal diagnosis. (a) Axial CTscan of upper portion of abdomen of patientpresenting with diffuse abdominal pain showsthrombosis of superior mesenteric vein (ar-row). (b, c) Axial CT scans show slightlythickened appendix (arrow) with periappendi-ceal stranding, without contrast material ex-travasation. Stopping the search for other ab-normalities after finding superior mesentericvein thrombosis (seen in a) could have beena pitfall in the diagnosis of appendicitis.(d, e) Axial CT scans obtained more ceph-alad show mild periappendiceal stranding (ar-row) where thrombosis of superior mesen-teric vein is intimately related.

CT in

Acu

te Ab

do

min

al Disease

125

tions of the method of examination. Most diverticularabscesses are quickly walled off and confined, but freeperforation, with pus and air in the peritoneal cavity,may occur. Diffuse peritonitis also occurs because ofperitoneal fluid dynamics and the interrelationshipsof the abdominal viscera.

Diverticular abscess may be mistaken for a mass(Fig 10) or an unopacified segment of colon (18–20).Colonic underfilling, spasm, and intraluminal contentcan make it difficult to determine if the colonic wall isgreater than 4 mm thick. Unfamiliarity with the ap-pearance of inflamed bowel wall from various condi-tions and the noninflammatory causes of thickening,such as submucosal fat, may cause perceptive errors.Identification of the inflamed diverticulum and asso-

ciated inflammatory stranding increases the specificityof imaging diagnosis.

Difficulty in the differentiation of diverticulitis fromcolon carcinoma is a major interpretive limitation(21–24). Colon carcinoma will perforate in about 5%of the patients, resulting in inflammatory changesusually characteristic of diverticulitis. The combina-tion of a length of involved colonic segment greaterthan 10 cm and the presence of mesenteric vascularengorgement favors a diagnosis of diverticulitis. Thepresence of adenopathy and a shelflike margin of thecolonic thickening favors a diagnosis of colon carci-noma. However, no single finding is 100% specific(21–24). In a group of examinations with similarnumbers of patients with diverticulitis and patients

Figure 10. Interpretive error in differentialdiagnosis of diverticulitis versus colon can-cer. (a, b) Axial CT scans of pelvis inmiddle-aged patient with anemia and lowerabdominal pain show stranding in sigmoidmesocolon and (b) eccentric thickening ofinferior margin of sigmoid suggestive of car-cinoma. Colonoscopy performed after 2weeks of antibiotic treatment revealed mul-tiple diverticula in sigmoid, without evidenceof mass.

Figure 9. Interpretive pitfall from mislead-ing clinical history in patient suspected ofhaving infection of aortoiliac graft. (a) AxialCT scan obtained at level of graft showsround area of soft-tissue attenuation (arrow).Perigraft thickening with small amount of gasis suggestive of graft infection. (b–e) Suc-cessive axial CT scans obtained inferior toscan in a show gas-filled appendix (arrow).At surgery, tip appendicitis localized to lat-eral graft limb was found. Findings frompathologic examination confirmed appendixadjacent to graft, with extensive transmuralinflammation, mucosal ulceration, and fat ne-crosis in mesoappendix.

Mag

linte

et a

l

126

with colon carcinoma, combined findings were shownto allow a confident diagnosis of diverticulitis in onlyabout 40% of the patients with diverticulitis and 66%of the patients with colon carcinoma (21). In at least10% of the patients coming to the emergency depart-ment, a confident diagnosis cannot be made. Failure tonote colonic wall thickening in a collapsed colon anddifficulty in differentiating muscular wall hypertrophyfrom colonic wall edema are interpretive pitfalls andmay result in diagnostic misclassification. Diverticulitisis a less common cause of colon obstruction than is co-lon carcinoma.

Recognition of an alternative diagnosis is particularlyimportant in diverticulitis because these patients tendto be older than patients with appendicitis and have agreater variety of alternative diagnoses. Radiologistsshould be familiar with the CT findings of primaryepiploic appendagitis, or omental torsion, which canmimic diverticulitis (Fig 11). Secondary epiploicappendagitis may complicate diverticulitis.

The techniques for evaluation of patients suspectedof having diverticulitis are variable. However, the usualtechnique consists of using both oral and intravenouscontrast materials. Sometimes an iodinated rectal con-trast material or air is added in puzzling cases, and thecolon has to be filled to assess colonic wall thickness.CT section thickness greater than 5 mm may obscurethe subtle fat stranding seen in mild diverticulitis. Useof thick CT sections may account for the high false-negative rate reported by early investigators. The use of

neutral (water) colonic contrast material with intrave-nous contrast enhancement needs to be investigated.

Intestinal Ischemia

Mesenteric ischemia or infarction occurs in a varietyof conditions that result in interruption or reductionof the blood supply of the intestine. Regardless of thecauses of the ischemic insult, the end results are simi-lar and range from transient alteration of bowel activityto transmural hemorrhagic necrosis. Mesenteric is-chemia is classified into four categories that are distinctconditions with different causes, clinical manifesta-tions, therapy, and prognoses. These four categories are(a) acute mesenteric ischemia, an acute ischemia ofthe small bowel with or without colonic involvement;(b) focal mesenteric ischemia, an acute ischemia of lo-calized segments of small intestine exemplified bystrangulating small-bowel obstruction; (c) chronic me-senteric ischemia, or ischemia without loss of tissue vi-ability; and (d) colonic ischemia. Early diagnosis is cru-cial because critical intestinal ischemia progresses to fa-tal infarction unless promptly diagnosed and treated.

The initial signs and symptoms of acute mesentericischemia are often nonspecific. Patients at risk includethose with the following: a history of prior mesentericischemia, vasculopathy, atrial fibrillation, nonthera-peutic anticoagulation therapy, hypotension (trauma,sepsis, cardiogenic shock), a history of recent myocar-dial infarction, congestive heart failure, or hypercoagu-lable states (cancer, etc) (25). Intestinal infarction ac-

Figure 11. Interpretive pitfallsecondary to indeterminate CTscan. Axial CT scans of patientwith right lower abdominal painshow stranding and mildly thick-ened small-bowel wall. Appen-dix or diverticula are not seen.CT scans were interpreted asequivocal for appendicitis orright-sided diverticulitis. At sur-gery, omental torsion was found.Appendix showed lymphoidhyperplasia.

Figure 12. Interpretive and perceptiveerror in diagnosis of intestinal ischemia.(a) Axial CT scan of elderly patient with acuteabdominal pain. Note dilated loops of smallbowel (arrows). Diagnosis was ileus. (b) Theclinical significance of extensive atheroscle-rotic calcification of superior mesenteric ar-tery (arrows) was not appreciated. Infarctedsmall bowel was found at surgery.

CT in

Acu

te Ab

do

min

al Disease

127

counts for 1% of the patients presenting with an acuteabdomen. Acute mesenteric ischemia is a life-threaten-ing condition, with mortality rates that range from59% to 93% (26). Findings from studies have shownthat early diagnosis and treatment have a substantialeffect on the outcome in a patient. Abdominal CT hasbeen considered of limited use in the diagnosis ofacute mesenteric ischemia, except in patients sus-pected of having superior mesenteric vein thrombosis(26). Perceptive errors in the diagnosis of venousthrombosis result from focusing on a more commonabnormality that explains the clinical manifestation.

Specific findings of bowel ischemia are relativelyuncommon in patients evaluated with CT. As such,radiologists are most often required to rely on second-ary CT findings, such as bowel wall thickening, pneu-matosis intestinalis, portomesenteric venous gas, orascites, as indirect signs of bowel ischemia. Any ofthese findings alone or in combination may raise sus-picion for this diagnosis, but they can be subtle andlack specificity, particularly in the absence of a clinicalmanifestation suggestive of ischemia (27–31). Thesefindings result in interpretive errors (Fig 12). Al-though these secondary findings are important in theCT evaluation and clinical management of acute ab-dominal pain, the pitfalls of CT imaging merit specialconsideration. Bowel ischemia remains one the mostdifficult diagnoses to establish with CT imaging crite-ria alone. Unlike many other diagnoses for which CTfindings can be pathognomonic, considerable empha-sis must be placed on the need for appropriate clinicalcorrelation of CT findings to establish a diagnosis ofbowel ischemia and guide appropriate further work-up and management.

The nonspecificity of clinical findings, the constella-tion of nonspecific CT findings, and the limitation ofCT technology prior to the use of multi–detector rowCT technology account for most of the errors reportedin the literature in the diagnosis of early intestinal is-chemia (Fig 13). Lack of optimization (timing, rate)

of the intravenous contrast material bolus, thick col-limation, single-phase acquisition, and the failure toachieve adequate intestinal opacification contribute totechnical pitfalls in the diagnosis of intestinal ischemia.Findings noted at angiography involving small vesselswere difficult to depict with single-detector helical CTtechnology. In addition, patients are rarely brought tothe radiology department during an acute episode ofischemia, so nonspecific CT findings of mucosal en-hancement from reperfusion are shown, rather thanthe more specific finding of intestinal ischemia: lackof mucosal enhancement from arterial insufficiency(32). Lack of familiarity with mesenteric vascular ana-tomic structures on cross-sectional imaging results inperceptive errors.

In patients with intestinal obstruction, the recogni-tion of complicating ischemia in the absence of a rec-ognizable closed-loop configuration may be difficult.Mesenteric fat stranding and ascites have a low sensi-tivity. The presence of interloop mesenteric fluid in-creases both sensitivity and specificity. When two orthree of these findings are present, the CT specificityfor strangulated obstruction is high (94%).

Although the imaging features of intestinal ischemiaare relatively nonspecific, an understanding of thepathophysiologic function and clinical features of thisdisease in various conditions and the radiologic find-ings allow the radiologist to consider ischemic boweldisease and its differential diagnosis and arrive at acorrect diagnosis (Fig 14).

For every complex problem, there is a simple solution. Andit’s always wrong.

—H. L. Mencken (33)

COMMENTS AND RECOMMENDATIONS

Precise categorization of errors in the performanceand interpretation of CT of the acute abdomen isdifficult because of the contribution of many factors.

Figure 13. Interpretive errors secondary tononspecific imaging findings. (a) Diffusethickening of long segments of pelvic loopsof ileum in patient with acute abdominal painsimulates CT findings of small-bowel Crohndisease. Colonoscopy revealed changesconsistent with ischemia. (b) Pelvic CT scanobtained in another patient with severe lowerabdominal pain shows rectal wall thickening(arrows) and perirectal stranding (arrow-heads), which were ascribed to inflammatorydisease. Focal ischemia seen at proctos-copy was not considered.

Mag

linte

et a

l

128

However, errors can be decreased by understandingcommon pitfalls, optimizing CT technique, and per-forming additional maneuvers to increase the accuracyof interpretation. Acknowledged inherent limitationsshould be recognized, and more sensitive methods ofexamination in the further work-up of patients shouldbe recommended to prevent repeated performance ofexaminations with low sensitivity.

Indeterminate scans caused by unfilled loops ofsmall bowel, appendix, or colon can be eliminated byadministering oral or rectal contrast material. Errorssecondary to technical limitation or “missed” lesionsthat are beyond the capability of the imaging exami-nation to demonstrate as clinical pathologic findingsare often related to lack of distention or an absence ofenteral volume challenge, which results in a lower sen-sitivity of CT for lower grades of small-bowel obstruc-tion and precludes exclusion of small-bowel obstruc-tion in the appropriate clinical setting. These patientsoften come to the emergency department.

The results of a critical analysis of the accuracy of CTin the diagnosis of small-bowel obstruction have shownan overall accuracy of 65% (9). With high-grade partialor complete obstruction, however, the accuracy increasesto 81%. Unfortunately, with low-grade obstruction, theaccuracy is only 48%. You need to look for subtle cluesto justify recommending more sensitive methods of ex-amination. Findings of enteroparietal peritoneal adhe-sions are often present at conventional CT but are notperceived because of the lack of a transition zone in theabsence of adequate enteral volume–challenged exami-nations for low-grade obstruction. Distortion of the con-vex anterior margin of the bowel with loss of the fat

plane or thickening of the soft-tissue density betweenthe small bowel wall and the anterior parietal or poste-rior parietal peritoneum can be recognized, although nointraluminal pressure gradient may be present.

Thus, in the setting of a history of prior abdominalsurgery, clinical correlation for recurrent small-bowelobstruction can be suggested, and more sensitive en-teral volume–challenged examination of the smallbowel can be performed electively (Fig 1). Factors thatinfluence nonfilling of small-bowel loops and postop-erative and congenital anatomic alterations should berecognized. An understanding of the influence of peri-stalsis and retained fluid and digestive secretions in pro-ducing pseudo–bowel wall thickening and pseudo-masses should lead to suggesting more appropriatemethods of investigation.

Recognizing normal structures, such as the right colicartery, that can mimic a normal appendix will decreasefalse-negative interpretations. A failure to opacify or rec-ognize the appearance of a normal cecum and appen-dix results in perceptive and interpretive errors that canbe avoided by optimizing CT technique.

Although there is now consensus that appendiceal CTshould include thin-section scanning of the right lowerquadrant of the abdomen, disagreement still persists re-garding the need for intravenous, oral, or rectal contrastmaterial (34–36). The most conservative and conven-tional approach is to perform helical CT of the abdo-men and pelvis with both intravenous and oral contrastmaterial. This is the most popular method and has al-lowed alternative diagnosis of appendicitis. Intravenouscontrast material has been shown to aid the diagnosisof appendicitis by facilitating identification of the in-

Figure 14. Interpretive error secondaryto nonspecific findings. (a) Axial CT scanof upper portion of abdomen of elderly manwith known metastatic prostatic cancer whopresented with acute abdominal pain. Portalvenous gas is seen. Also note metastatic fociin liver. (b) Axial CT scan at level of gallblad-der shows gas in portal vein. (c) Axial CTscan at level of midportion of abdomenshows dilated small bowel and "misty" mes-entery. (d) Axial CT scan at level of kidneysshows dilated small bowel displacing col-lapsed ascending colon. Note atheroscle-rotic plaque in abdominal aorta. At surgery,there was no evidence of intestinal infarc-tion. Small-bowel obstruction was from ad-hesions, which were lysed. Patient had anuneventful postoperative course. (Imagescourtesy of Stefania Romano, MD, Naples,Italy.)

CT in

Acu

te Ab

do

min

al Disease

129

flamed appendix in those who have mild forms of ap-pendicitis, in whom the diagnosis may rest solely onidentification of luminal dilatation and pathologic ap-pendiceal wall enhancement (13,37–39). Use ofnonenhanced helical CT of the abdomen and pelviswithout intravenous, oral, or rectal contrast materialhas been proposed because it can be done faster with-out exposing patients to iodinated contrast material(40,41). This technique may be effective in patientswith large body habitus. Proponents of oral contrastmaterial advocate its use to limit misinterpretation ofmimics of appendicitis and to allow diagnosis of small-bowel diseases (34).

Because of the length of time needed to opacify theileocecal region, a focused appendiceal CT protocolhas been proposed (42,43). This protocol entails ad-ministration of colonic contrast material and acquisi-tion of a limited helical CT study of the right lowerquadrant. This technique has been shown to be as ac-curate as other protocols, and it increases the negativepredictive value of CT in the diagnosis of acute appen-dicitis and can be completed within 15 minutes in themajority of patients.

Clearly, agreement has not been reached on an opti-mized technique to address all of the pitfalls dis-cussed. The interested reader can look at the technicaldetails of these different methods.

Because there is controversy in the radiology litera-ture with regard to the use of oral contrast material,some radiologists are now exploring the use of water ora lower-density iodinated or barium contrast materialas the oral contrast material, along with intravenouscontrast enhancement, to use the capabilities of multi–detector row CT technology to show mucosal enhance-ment, which can be masked by the intraluminal opac-ity of currently used oral radiopaque contrast material(44). Early experience has shown that a high volume ofwater as oral contrast material, coupled with intrave-nous contrast enhancement, appears to offer a feasible

substitute for radiopaque oral contrast material forevaluating the small bowel (45). The use of neutral oralcontrast material (water) with intravenous contrast ma-terial, instead of radiopaque oral contrast material, mayprevent admixture defects and pseudo–bowel wallthickening and allows recognition of subtle mucosalhyperemia and submucosal edema. Mucosal hyper-emia may be isoattenuating relative to the attenuationof the currently used radiopaque oral contrast materialin the lumen of the small bowel but can be seen withneutral (water) or a lower-density radiopaque contrastmaterial. Not infrequently, patients presenting withacute abdominal pain have nausea and vomiting,which could be aggravated by the bad taste of the cur-rently used radiopaque oral contrast materials.

On the basis of personal clinical experience per-forming CT enteroclysis with neutral (water or meth-ylcellulose) enteral contrast materials combined withintravenous contrast enhancement, an abdominal-pelvic CT protocol is suggested (Fig 15). With thisprotocol, the study should be carefully monitored,and multi–detector row CT parameters should beproperly selected to tailor the examination to theclinical query, particularly if mesenteric ischemia is aleading clinical possibility. This protocol should beaugmented with a focused appendiceal CT protocolwhen imaging findings are equivocal or anatomiclimitations (no fat planes) exist for reliable diagnosisof acute appendicitis in a patient in whom appendici-tis was not the primary diagnostic consideration. Inthe imaging of the true acute abdomen, appendicitis isnot ruled out if the normal appendix is not depictedor filled with contrast material from its base to the tip.

The unfounded fear that water does not distend dis-tal small-bowel loops is due to administration of in-adequate amounts of water and the lack of adminis-tration of a hypotonic agent prior to scanning. Fur-thermore, the initial use of a peristaltic agent ensuresfilling of distal small-bowel loops and colon.

The diagnosis of early intestinal ischemia will requireclose communication between emergency physiciansand radiologists. In patients who present with an acuteabdomen and a clinical history of one or more of thehigh-risk factors for acute mesenteric ischemia, the find-ings from early experience indicate that biphasic CT withmesenteric CT angiography with the use of multi–detec-tor row CT technology appears to be effective in the diag-nosis of acute mesenteric ischemia (32). Further experi-ence is needed with this method of examination.

Optimizing CT technique to address all pitfalls de-scribed will undoubtedly decrease errors in the CT di-agnosis of the acute abdomen. No single techniquecan address all of the pitfalls and limitations. Localpractice will influence how the CT technique will beoptimized. To take advantage of multi–detector rowCT technology, a method that results in luminal disten-tion and opacification of the entire small bowel and

Figure 15. Abdominal/pelvic CT protocol that uses water asoral contrast material. Instead of water, commercially preparedbottled flavored drinks (or such drinks diluted in water) or very-low-density oral contrast material may be substituted.

Mag

linte

et a

l

130

colon, combined with intravenous contrast enhance-ment, should be adopted.

A bowel-focused technique has been described, aminor variation to the routine abdominal scanning,geared toward a more substantial bowel filling with oralcontrast material and use of multiplanar reformatting.This technique has been termed “CT enterography” byRaptopoulos et al (46,47). CT enterography also ad-dresses more than small-bowel abnormalities.

In CT enterography, approximately 1000–1500 mLof a 2% barium-based or 2%–2.5% water-soluble io-dine-based oral contrast material is administered dur-ing a period of 1–2 hours before scanning. A high doseof intravenous contrast material and a biphasic injec-tion regimen have been recommended. First, 30–50 mLof contrast material is administered at 2 mL/sec with-out CT acquisition. After a delay of 2–3 minutes, the re-mainder of the 150 mL of solution containing 300 mgof iodine per milliliter is given. Eighty to one hundredmilliliters is infused at 2–3 mL/sec. Scanning starts 60–70 seconds after the second dose. This method results invascular opacification, parenchymal enhancement, andcombined renal parenchymal and excretory phase scan-ning (46). CT parameters will depend on the vendorand model of multi–detector row CT scanner. Refor-matting may be done on a dedicated workstation. An-other bowel-focused method that has been describeduses a neutral enteral contrast material, polyethyleneglycol and whole milk, as well as an isotonic oral solu-tion (48,49).

The volumes of oral contrast material required inthese alternative methods may not be feasible to usein a nauseated patient, and the length of time neededto ingest the recommended volume may be impracti-cal in a busy emergency department, where speed isrequired in the performance of these examinations.These methods can be readily used if a patient has anasogastric tube. The methods that use neutral enteralcontrast material (water, polyethylene glycol withmilk) together with biphasic CT acquisition should beof value in patients suspected of having intestinal is-chemia. The techniques that use neutral oral and en-teral contrast material need to be supplemented witha focused appendiceal CT examination when appendi-citis as a possibility has not been ruled out. None ofthese newer alternative methods have been validatedin a large group of emergency department patients.The use of a large volume of water with initial use of apropulsive medication (metoclopramide) can shortenthe lag time of some of the proposed methods. Use ofa hypotonic agent allows partial distention of small-bowel loops. Admixture defects and pseudomasseswill be prevented by the use of a neutral enteral con-trast material, which also allows depiction of mucosalenhancement.

Obtaining more clinical information than the usualcomplaint of “abdominal pain” or “acute abdomen”

enhances the diagnostic utility of CT because the clinicalinformation provided on the imaging requests is fre-quently inadequate. In most imaging requests for emer-gent CT for acute abdomen indications, the condition israrely a true acute abdomen. Current advances in infor-mation technology allow radiologists to immediatelyretrieve the clinical background data of a patient.

In conclusion, the past several years have spannedan ongoing revolution in CT technology. Improve-ment in technology led to the development of single-detector scanning, followed by multi–detector row CTscanning. Advanced computer workstations with spe-cial hardware are now available to process the largevolumes of data that the newest multi–detector rowscanners produce (44).

Optimizing the protocol for abdominal CT will bethe key to addressing diagnostic pitfalls and limitationsbut may not be possible with a single imaging method.Emergency radiologists must use the advantages ofmulti–detector row CT technology and understand thereasons for the various pitfalls and limitations. Carefulmonitoring of the CT examination and the use of addi-tional techniques to opacify the appendix when it isnot completely depicted will diminish errors in the di-agnosis of acute appendicitis. Peritonitis from perfo-rated appendicitis is a great mimic of small-bowel dis-ease. Misdiagnosis of appendicitis tends to occur in pa-tients with paucity of intraabdominal fat. Pseudo-obstruction from acute appendiceal perforation, ileusfrom peritoneal irritation, or hypoperistalsis frommedication can lead to a false-positive diagnosis ofsmall-bowel obstruction.

The attention of the radiologist should not be di-verted by a misleading clinical history or atypical clini-cal manifestation. Retrieval of patient information,which is available with advances in information tech-nology, should be used more often. The best allies ofthe radiologist are experienced emergency physiciansand general surgeons who examine their patients firstbefore sending them to the radiology department forabdominal CT. Close clinical communication withthese physicians will allow radiologists to tailor ab-dominal CT protocols by using multi–detector row CTtechnology to address the precise clinical query.

It’s what you learn after you know it all that counts.—John Wooden, Hall of Fame basketball coach

(50)

References

1. Roosevelt E. Quoted by: Levine R. The power of persuasion:how we’re bought and sold. Hoboken, NJ: Wiley, 2003; 4.

2. Maglinte DD, Rubesin S. Advances in intestinal imaging: pref-ace. Radiol Clin North Am 2003; 41:xi–xii.

3. Maglinte DD, Burney BT, Miller RE. Lesions missed on small-bowel follow-through: analysis and recommendations. Radiol-ogy 1982; 144:737–739.

4. Jaffe CC. The meaning of a “negative” examination (editorial).AJR Am J Roentgenol 1980; 134:414–415.

CT in

Acu

te Ab

do

min

al Disease

131

5. Sitting KM, Korr MS, McDonald JC. Abdominal wall, umbilicus,peritoneum, mesenteries, omentum, and retroperitoneum. In:Sabiston DC Jr. Textbook of surgery: the biological basis ofmodern surgical practice. 15th ed. Philadelphia, Pa: Saunders,1996; 809–823.

6. Meyers M. Dynamic radiology of the abdomen: normal andpathologic anatomy. 4th ed. New York, NY: Springer, 1994.

7. Maglinte DD, Kelvin FM, O’Connor K, Lappas JC, Chernish SM.Current status of small bowel radiography. Abdom Imaging1996; 21:247–257.

8. Maglinte DD, Balthazar EJ, Kelvin FM, Megibow AJ. The role ofradiology in the diagnosis of small-bowel obstruction. AJR Am JRoentgenol 1997; 168:1171–1180.

9. Maglinte DD, Gage SN, Harmon BH, et al. Obstruction of thesmall intestine: accuracy and role of CT in diagnosis. Radiology1993; 188:61–64.

10. Herlinger H, Maglinte DD. Small bowel obstruction. In: HerlingerH, Maglinte DD, eds. Clinical radiology of the small intestine.Philadelphia, Pa: Saunders, 1989; 479–507.

11. Sarr MG, Bulkley GB, Zuidema GK. Preoperative recognition ofintestinal strangulation: prospective evaluation of diagnostic ca-pability. Am J Surg 1983; 145:176–182.

12. Bendeck SE, Nino-Murcia M, Berry GJ, Jeffrey RB Jr. Imagingfor suspected appendicitis: negative appendectomy and perfo-ration rates. Radiology 2002; 225:131–136.

13. Raptopoulos V, Katsou G, Rosen MP, Siewert B, Goldberg SN,Kruskal JB. Acute appendicitis: effect of increased use of CT onselecting patients earlier. Radiology 2003; 226:521–526.

14. Rao PM. Technical and interpretative pitfalls of appendiceal CTimaging. AJR Am J Roentgenol 1998; 171:419–425.

15. Callahan MJ, Rodriguez DP, Taylor GA. CT of appendicitis inchildren. Radiology 2002; 224:325–332.

16. Urban BA, Fishman EK. Targeted helical CT of the acute abdo-men: appendicitis, diverticulitis, and small bowel obstruction.Semin Ultrasound CT MR 2000; 21:20–39.

17. Rhea JT. CT evaluation of appendicitis and diverticulitis. I. Ap-pendicitis. II. Diverticulitis. Emerg Radiol 2000; 7:237–244.

18. Akbar SA, Shirkhoda A, Jafre SZ. Pictorial essay: normal vari-ants and pitfalls in CT of the gastrointestinal and genitourinarytracts. Abdom Imaging 2003; 28:115–128.

19. Macari M, Balthazar EJ. CT of bowel wall thickening: significanceand pitfalls of interpretation. AJR Am J Roentgenol 2001;176:1105–1116.

20. Wittenberg J, Harisinghani MG, Jhaveri K, Varghese J, MuellerPR. Algorithmic approach to CT diagnosis of the abnormalbowel wall. RadioGraphics 2002; 22:1093–1109.

21. Chintapalli KN, Chopra S, Ghiatas AA, Esola CC, Fields SF,Dodd GD III. Diverticulitis versus colon cancer: differentiationwith helical CT findings. Radiology 1999; 210:429–435.

22. Balthazar EJ, Megibow A. Schninella RA, et al. Limitations inthe CT diagnosis of acute diverticulitis: comparison of CT, con-trast enema, and pathologic findings in 16 patients. AJR Am JRoentgenol 1990; 154:281–285.

23. Hulnick DH, Megibow AJ, Balthazar EJ, Gordon RB, SurapeniniR, Bosniak MA. Perforated colorectal neoplasms: correlation ofclinical, contrast enema, and CT examinations. Radiology 1987;164:611–615.

24. Van Breda Vriesman AC, Puylaert JB. Pictorial essay: epiploicappendagitis and omental infarction: pitfalls and look-alikes.Abdom Imaging 2002; 27:20–28.

25. Wiesner W, Khurana B, Ji H, Ros PR. CT of acute bowel is-chemia. Radiology 2003; 226:635–650.

26. Brandt LJ, Boley SJ. AGA technical review on intestinal is-chemia: American Gastrointestinal Association. Gastroenterol-ogy 2000; 118:954–968.

27. Bartnicke BJ, Balfe DM. CT appearance of intestinal ischemiaand intramural hemorrhage. Radiol Clin North Am 1994; 32:845–860.

28. Frager D, Baer JW, Medwid SW, et al. Detection of intestinalischemia in patients with acute small bowel obstruction due

to adhesions or hernia: efficacy of CT. AJR Am J Roentgenol1996; 166:67–71.

29. Balthazar EJ, Liebeskind ME, Macari M. Intestinal ischemia inpatients in whom small bowel obstruction is suspected: evalua-tion of accuracy, limitations, and clinical implications of CT in di-agnosis. Radiology 1997; 205:519–522.

30. Sebastia C, Quiroga S, Espin E, Boye R, Alvarez-Castells A,Armengol M. Portomesenteric vein gas: pathologic mechanisms,CT findings, and prognosis. RadioGraphics 2000; 20:1213–1224.

31. Rha SE, Ha HK, Lee SH, et al. CT and MR imaging findings ofbowel ischemia from various primary causes. RadioGraphics2000; 20:29–42.

32. Kirkpatrick IDC, Kroeker MA, Greenberg HM. Biphasic CT withmesenteric angiography in the evaluation of acute mesentericischemia: initial experience. Radiology 2003; 229:91–98.

33. Mencken HL. Quoted by: Levine R. The power of persuasion:how we’re bought and sold. Hoboken, NJ: Wiley, 2003; 137.

34. Birnbaum BA, Wilson SR. Appendicitis at the millennium. Radi-ology 2000; 215:337–348.

35. Birnbaum BA, Jeffrey RB Jr. CT and sonographic evaluation ofpatients with acute right lower quadrant pain. AJR Am JRoentgenol 1998; 170:361–371.

36. Weltman DI, Yu J, Krumenacker J Jr, Huang S, Moh P. Diag-nosis of acute appendicitis: comparison of 5- and 10-mm CTsections in the same patient. Radiology 2000; 216:172–177.

37. Raman SS, Lu DS, Kadell BM, et al. Accuracy of nonfocusedhelical CT for the diagnosis of acute appendicitis: a 5-year re-view. AJR Am J Roentgenol 2002; 178:1319–1325.

38. Jacobs JE, Birnbaum BA, Macari M, et al. Acute appendicitis:comparison of helical CT diagnosis—focused technique with oralcontrast material versus nonfocused technique with oral and in-travenous contrast material. Radiology 2001; 220:683–690.

39. Balthazar EJ, Birnbaum BA, Yee J, Megibow AJ, Roshkow J,Gray C. Acute appendicitis: CT and US correlation in 100 pa-tients. Radiology 1994; 190:31–35.

40. Lane MJ, Katz DS, Ross BA, Clautice-Engle TL, Mindelzun RE,Jeffrey RB Jr. Unenhanced helical CT for suspected acute ap-pendicitis. AJR Am J Roentgenol 1997; 168:405–409.

41. Lane MJ, Liu DM, Huynh MD, Jeffrey RB Jr, Mindelzun RE, KatzDS. Suspected acute appendicitis: nonenhanced helical CT in300 consecutive patients. Radiology 1999; 213:341–346.

42. Rao PM, Rhea JT, Novelline RA, et al. Helical CT technique forthe diagnosis of appendicitis: prospective evaluation of a focusedappendix CT examination. Radiology 1997; 202:139–144.

43. Rao PM, Rhea JT, Novelline RA, Mostafavi AA, Lawrason JN,McCabe CJ. Helical CT combined with contrast material admin-istered only through the colon for imaging of suspected appen-dicitis. AJR Am J Roentgenol 1997; 169:1275–1280.

44. Horton KM, Fishman EK. The current status of multidetectorrow CT and three-dimensional imaging of the small bowel.Radiol Clin North Am 2003; 41:199–212.

45. Wold PB, Fletcher JG, Johnson CD, Sandborn WJ. Assessmentof small bowel Crohn disease: noninvasive peroral CT enterog-raphy compared with other imaging methods and endoscopy—feasibility study. Radiology 2003; 229:275–281.

46. Raptopoulos V, Schwartz RK, McNicholas MM, Movson J,Pearlman J, Joffe N. Multiplanar helical CT enterography in pa-tients with Crohn’s disease. AJR Am J Roentgenol 1997; 169:1545–1550.

47. Rosen MP, Siewert B, Sands DZ, Bromberg R, Edlow J,Raptopoulos V. Value of abdominal CT in the emergency de-partment for patients with abdominal pain. Eur Radiol 2003;13:418–424.

48. Thompson SE, Raptopoulos V, Sheiman RL, McNicholas MM,Prassopoulos P. Abdominal helical CT: milk as a low-attenuationoral contrast agent. Radiology 1999; 211:870–875.

49. Mazzeo S, Caramella D, Battola L, et al. Crohn disease of thesmall bowel: spiral CT evaluation after oral hyperhydration withisotonic solution. J Comput Assist Tomogr 2001; 25:612–616.

50. Wooden J. Quoted by: Maxwell JC. The 21 indispensable quali-ties of a leader. Nashville, Tenn: Thomas Nelson, 1999; 141.

NO

TES

133

The Contemporary Role ofConventional Radiographs

in Evaluating theAcute Abdomen1

The last RSNA Categorical Course in Emergency Radiology took place at the RSNAscientific assembly in 1995. In the past 9 years, the imaging approach to the emergentlyill patient with abdominal discomfort has changed radically. The capabilities of ultra-sonography (US) were clearly recognized back then, but its current role has really notexpanded much. Developments in magnetic resonance imaging, both instrumentaland interpretive, offer promise, yet this technique still remains on the periphery. Therevolution in emergent abdominal imaging—and it has been a revolution—is encom-passed largely by developments in computed tomographic (CT) technology and itsrapid and largely uncritical embrace by emergency physicians, surgeons, internists,and radiologists alike. A dependence on CT has caused CT findings to displace, to alarge degree, the clinical history, the findings from physical examination, and evenlaboratory data as the first data to be assessed in the quest for a diagnosis in patientswith acute abdominal disease.

Where does this dependence on CT leave the conventional radiograph? Is it just avenerable, but limited, image characterized as a relic of a bygone era? Is the conven-tional radiograph maintained as an exercise only because of the impress of tradition,obtained as a matter of course but considered fundamentally irrelevant as our atten-tion is promptly and inevitably directed to CT? Should we term the performance ofconventional radiography merely a vestigial ritual having really no place in our arma-mentarium that provides any value to the patient?

The answer to this challenge, I believe, is tripartite: (a) At present, for two com-mon acute conditions, conventional radiography does have an essential role. (b) Inthe medium term, a synthesis of conventional radiography and CT may occur, if re-imbursement incentives are recast. (c) In the long run, a renewal of interest willlikely occur as the issue of excessive radiation dose, especially in children and youngadults, takes its place on the national agenda, where not only physicians but alsohealth policy makers and the public at large all lend their strident voices to the de-bate about risk, cost, and benefit.

LIMITATIONS OF USE

First things first! The conventional radiograph is important today for only two catego-ries of nontraumatic acute disease evaluated at an emergency facility: pneumoperito-neum and intestinal obstruction. For both, the nonenhanced supine radiograph can

Stephen R. Baker, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 133–141.

1From the Department of Radiology, University Hospital, New Jersey Medical School, PO Box 1709, Room C318, 150Bergen St, Newark, NJ 07101 (e-mail: bakersr@umdnj.edu).

Bak

er

134

often provide the definitive diagnostic informationwith a sensitivity equal to that of CT. Although thenonenhanced spine radiograph cannot reveal the wel-ter of additional spatial relationships that are the hall-mark of a CT study, it can demonstrate the key findingrequiring undelayed medical therapy or surgical inter-vention. Almost always, the observation of bowel ob-struction on conventional radiographs should engen-der an undelayed clinical response. Likewise, the recog-nition of free air cannot be ignored. In most instancesof these two emergencies, supplementary findings dis-cernible with CT will most often not be crucial for theinitiation or modification of immediate therapy.

Conventional abdominal radiographs can some-times allow detection of a tumefaction, if the mass islarge enough to displace stomach or bowel gas or toobliterate fat planes. Yet the plain films cannot delin-eate detailed information about the precise location,exact size, and particularities of tumor contour withany degree of precision. Conventional radiographycan be used to discern the small bubbles of pus collec-tions, but a comprehensive display of the extent, con-figuration, and multiplicity of abscesses is beyond itscapabilities. Most abdominal calcifications are observ-able on conventional radiographs, and the differentialdiagnosis can almost always be limited to no morethan two or three entities with an evaluation of themultiplicity, size, shape, internal architecture, location,and movement of such calcifications on successiveimages. However, the effect of the calcification on thelumen in which is contained or on the structures thatthe calcification abuts requires further investigationwith contrast agent–enhanced radiography, US, or CT.

It is important, then, to know much about the con-ventional radiographic appearances of free air andbowel obstruction because even today, conventionalradiography may be the only study available to detectthese conditions in certain facilities and at certaintimes of the day. Moreover, it is critical for radiolo-gists to maintain an interest in the intricacies of con-ventional radiographic interpretation because it islikely that in the future, CT use will be regulated moreclosely and is apt to be restricted in utilization tominimize dose accumulation—a practice that existstoday in many countries in the European Union.

RELATIVE VALUE OF HISTORY

It is a given that knowledge of the clinical history is vi-tal, not just for emergency patients but for all manifes-tations, before a diagnosis that is based on imagingstudies can be made. That having been said, once ithas been made clear that the possible presence ofpneumoperitoneum is not related to a recent abdomi-nal operation or biopsy and is not a consequence ofthe rare situations in which air has been introducedinto the peritoneal space by purposeful human activ-

ity, then the history is actually not so important and, infact, can be misleading. Most patients who have a per-forated ulcer experience immediate and severe pain.On the other hand, perforation of the colon may ex-hibit a slower tempo in its generation of symptoms.Consequently, the radiographic signs of pneumoperito-neum may appear before substantial patient discomfortsupervenes. For the elderly and for demented individu-als or those receiving steroids, perforation of a hollowviscus, especially if it is the stomach or duodenum, isoften associated with no sensation of distress or painand can be accompanied by normal findings at physi-cal examination. Hence, the emergency radiologistshould examine every conventional radiograph of theabdomen with a high index of suspicion, searching forthe subtle manifestations of free air even though nocorroborating information is available to him or her.

On the contrary, the diagnosis of bowel obstructionunequivocally depends on the correlation of clinicaland radiographic findings. The radiologist who has at-tempted to render this diagnosis when no accompany-ing clinical information is available, either from thehistory, physical examination, or the acquisition oflaboratory data, will be inevitably placing the patientin jeopardy and placing himself or herself and the re-ferring physician at risk for a malpractice suit. The rea-son is that the conventional radiographic manifesta-tions of luminal occlusion can (a) often be simulatedby an adynamic ileus, (b) sometimes be mimicked bymesenteric ischemia without intestinal obstruction

Figure 1. Simulation of bowel obstruction by mesenteric is-chemia. Radiograph shows dilated small bowel and nondilatedcolon, caused by acute vascular insufficiency. There was noluminal occlusion.

Co

nven

tion

al Radio

grap

hs o

f the A

cute A

bd

om

en

135

(Fig 1), and (c) occasionally be imitated by air swal-lowing alone. Thus, the radiographic finding of dilata-tion of bowel segments proximal to the point of a pu-tative intestinal blockage must be considered within aparticular clinical context, not in isolation from thehistory, physical findings, and laboratory data.

THE FALLACY OF THE KIDNEY-URETER-BLADDER RADIOGRAPH

Early in the 20th century, the term KUB (for kidney-ureter-bladder radiograph) gained prominence andthen dominance as both an appellation and as the ac-cepted standard image for a full assessment of the ab-domen and pelvis. There are compelling reasons whythis limited projection with its prescriptive initials be-came the surrogate for a comprehensive radiographicinspection of the contents of that part of the body situ-ated below the diaphragm and down to the perineum.KUB was a name congruent with the particular interestsof urologists because it encompassed the span of theurinary tract. The KUB radiograph provided a suitableframe within which calculi could be detected.

Furthermore, the acceptance of the KUB radiographwas based on denial of an important photographic fact.For most American adults, with the use of a standard40-inch (102-cm) source-to-film distance for a supineradiograph, the entirety of the peritoneal cavity ex-tending from the uppermost hemidiaphragm rostrallyto the obturator foramina caudally cannot be encom-passed on a single radiograph. This fact has often leftthe upper reaches of the abdomen, which are situated

beyond the superior margin of the KUB radiograph,out of sight and, consequently, out of mind (Fig 2).The remedy for the deficiency of the KUB radiographwas to obtain upright radiographs of the chest or ab-domen and/or right-side-elevated decubitus views toassess the uppermost peritoneal cavity for signs of per-foration or obstruction. The assumption of an uprightposition is difficult for the inebriated patient, for thepatient with dementia, and especially for the weak pa-tient with obtundation. Moreover, the upright posi-tion is not necessary when CT is contemplated as theimminently available follow-up examination.

Nonetheless, a horizontal-beam erect radiograph ora decubitus erect radiograph, or both, remain as partof the customary abdominal series in many centers.However, the results of a recent investigation haveshown that upright and decubitus views are unneces-sary and should be replaced with a protocol involvingtwo slightly overlapping recumbent supine images,one of the upper part of the abdomen and the other ofthe middle part of the abdomen and pelvis (1).

MINIMAL DETECTABLE AMOUNT OF FREE AIR

In the early 1970s, Miller and co-workers (2,3), intheir landmark reports on detection of pneumoperito-neum, observed that as little as 1–2 mL of free aircould be seen on conventional radiographs when thesubject was placed in the upright or decubitus posi-tion. Miller made that observation by injecting ali-quots of air of varying volume into his own peritonealcavity. However, these investigations were constrainedby the same dogma that enabled the KUB radiographto gain ascendancy. These investigators presumed thatthe supine view was not sensitive for the detection offree air, inasmuch as it did not encompass the subdia-phragmatic space.

Today, with the presence of CT, we have reexaminedthose observations. By analogy alone, we know that anonenhanced radiograph is exquisitely able to depictgas bubbles anywhere in the abdomen, as evidenced bythe almost ubiquitous observation of tiny bubbles ofgas within intraluminal fecal deposits. Bubbles with di-ameters as small as 2 mm (giving them a volume of ap-proximately 0.004 mL) are routinely seen in the co-lonic shadow (Fig 3). If bubbles of gas can be discernedin the large bowel, then liberated gas should also berecognizable in the right upper quadrant overlying thehomogeneous gray shadow of the liver (1).

More recent attempts to determine the sensitivity ofvarious imaging approaches in addition to CT have in-cluded investigations of the upright abdomen, the lat-eral chest (4), the upright chest (5), US (6), and others.All of these approaches have two problems: (a) When apatient is repositioned from the supine orientation,bubbles of free air will move. (b) If the CT study is notdone simultaneously or at least contemporaneously

Figure 2. Anterior bubble sign (arrows), a manifestation offree air from a perforated anterior distal stomach ulcer. Radio-graph shows that the air is situated anterior to liver, well aboveright kidney.

Bak

er

136

with the performance of either conventional radiogra-phy or US, the CT image cannot be used as the refer-ence standard because more gas may be either liber-ated or absorbed in the interval between the two im-aging studies. However, the simple digital scout view,which is a part of almost every CT study of the abdo-men, provides the opportunity to determine the small-est amount of free air that could be seen on supine pro-jections, if that projection includes the uppermost partof the abdomen to encompass the otherwise featurelessgray background of the hepatic shadow (Fig 4). Thedigital scout view is obtained at the same time as thecross-sectional images, and the patient does not changeposition when both are obtained.

We have evaluated this issue with both phantomsand clinical examples. Our results indicate that in theupper part of the abdomen, the supine radiograph isjust as sensitive as the CT image for the detection ofsmall bubbles of free air (7,8). Consequently, we in-sist that an appropriately focused series consisting ofoverlapping supine radiographs be obtained to inter-rogate the entire abdomen. With such a protocol, anylucency in the right upper quadrant that is lateral andsuperior to the duodenal bulb and superior to the he-patic flexure must be explained before being dis-missed (Fig 5). Such an emphasis on the right upperquadrant has allowed us to recognize subtle signs offree air that otherwise would be missed with a routinesupine study consisting of only a KUB radiograph (9).

Moreover, the presence of free air in specific loca-tions over the liver shadow can enable an inference tobe drawn as to the likely location of the perforationfrom which the free air is derived. It takes approxi-mately 6–12 hours after perforation before a perito-

neal insult results in a generalized ileus (10). As men-tioned before, perforated ulcers in the stomach orduodenal bulb in a steroid-free, alert patient are gen-erally accompanied by the experience of immediatepain. On the other hand, colonic perforations aremore insidious in the development of signs andsymptoms. Therefore, in someone suspected of havingspontaneous pneumoperitoneum, the absence of anadynamic ileus is suggestive that the free air involvesthe stomach or duodenum, whereas the presence ofan adynamic ileus makes one look for a more distalperforation. Furthermore, free air in the right upperquadrant situated below the hemidiaphragm indicatesan anterior perforation, whereas the confinement ofair below the right 11th rib, while still intraperitoneal,is suggestive of localization in the Morison pouch, aposterior-superior intraperitoneal recess intimately re-lated to the posterior wall of the duodenal bulb.

MISCONCEPTIONS: CONVENTIONALRADIOGRAPHIC DIAGNOSIS OF INTESTINALOBSTRUCTION

In the 1950s, Dr Frimann-Dahl, an influential Norwe-gian radiologist, made a series of observations thatwere, in short order, widely accepted. His claims haveinfluenced pathophysiologic conceptions among sur-geons and radiologists for many decades now. The re-ports by Frimann-Dahl, which were based on fluoro-scopic examination of the abdomen, led to the notionof differential fluid levels as being a finding crucial tothe distinction of intestinal obstruction from ady-namic ileus. He stated that differing fluid levels in oneloop indicate obstruction, whereas fluid levels at the

Figure 3. Minimal pneumoperitoneum. (a) Digital scout view shows a sliver of free air (arrows) obliquely oriented above distal por-tion of stomach. (b) Corresponding transverse CT image shows a collection of free air (arrow) anterior to liver, conforming in site andorientation to the bubble seen on preliminary radiograph.

Co

nven

tion

al Radio

grap

hs o

f the A

cute A

bd

om

en

137

same height in a single bowel segment were evidenceof adynamic distention or ileus (11).

Frimann-Dahl made several assumptions that arenot correct. First, he stated that “at equilibrium,”such findings would be noted. However, there is no

such thing as equilibrium in the intestines becausethe peristaltic movement of air and liquid throughthe tubular gastrointestinal tract is never in a steadystate. The passive rising and falling of gas-liquid in-terfaces and the progression or reversal of peristalsisproximal to an obstruction mandate that luminalcontents are always in flux with respect to volumeand position. There is no quiet interval allowingfluid levels to equilibrate because fluid is constantlybeing introduced from above or returned above bymuscular contraction. Furthermore, liquid and gaswill sporadically pass through an intermittent ob-struction or will merely fill up an arm of one loopand start to drip into another arm of that same loopat any point in time. Thus, the equilibrium pro-pounded so compellingly by Frimann-Dahl does notexist in real life.

Moreover, Frimann-Dahl was making his observa-tions with fluoroscopic examination of a patient. Inessence, he was producing movies, while we takesnapshots either with CT or conventional radiogra-phy. CT has demonstrated clearly that interfaces ofdiffering heights can be observed in one loop, and atthe same time, fluid levels of the same height may beseen in either that same loop or an adjacent loop inpatients who have either mechanical obstruction oradynamic ileus. Therefore, the recognition of fluidlevels with horizontal-based radiographs, either withdecubitus or upright projections, does not advanceour understanding of intestinal obstruction (Fig 6).In addition, there is really no need to make patientsstand or lie on their side if the conventional radio-graphs are not diagnostic and if CT is contemplated.In corroboration of this skepticism, the findingsfrom a number of studies have shown the nonutilityof upright radiographs, in the presence of supine ra-diographs, for the determination of obstruction(11,12).

Figure 5. Intraperitoneal air. Radiograph shows bubbles (ar-rows) scattered over the liver, well above the colon and theduodenal bulb.

Figure 4. Inhomogeneity of the liver shadow caused by freeair. (a) Conventional radiograph shows areas of lucency (arrow)scattered over the hepatic shadow. (b) Transverse CT imageshows anterior free-air bubble (outlined in white).

Bak

er

138

BOWEL OBSTRUCTION: QUESTIONABLEASSUMPTIONS

A variety of false notions attend and confound a cor-rect conventional radiographic diagnosis of bowel ob-struction. Some have become fiercely held tenets thatare still readily accepted or at least acknowledged, yetthey fail to hold up with careful scrutiny.

Before proceeding with this list, some importantpoints of conventional radiographic interpretationshould be established. A guide to the location of thelarge bowel and, from that, to the observation of thestomach and small bowel depends on recognition ofthe appearance and position of the transverse colon.This intestinal segment is almost always available fordepiction because the colon almost always containssome gas. On supine views, the transverse colon isthe least dependent segment of the large bowel.Therefore, the transverse colon is most apt to be theplace where flatus will collect. Of course, the trans-verse colon (a) could have been removed in a prioroperation, (b) could be displaced upward or down-ward, depending on body habitus, and (c) may neverhave formed if the patient had a malrotated largebowel. However, in most cases, the transverse colon isreadily observable. After identifying the transverse co-lon, which is distinguishable by the presence of in-traluminal feces, by the bordering haustral outpouch-ings, and by the infoldings of the plicae semilunares,one should attempt to determine the location and thedegree of dilatation of the cecum (13).

A colonic landmark often sought, but rarely involvedin the determination of obstruction, is the rectal gas

shadow. In the presence of obstruction, the rectum maystill contain gas because the obstruction may be recentor intermittent or because an extensive amount of fecesbeyond the point of luminal occlusion may continueto ferment and produce flatus even when the obstruc-tion is total and unremitting. Thus, the presence of rec-tal gas does not mean that there is no obstruction. Con-versely, the absence of rectal gas does not mean thatthere is an obstruction. The rectum is the most depen-dent colonic segment, and therefore it is least apt tocontain flatus. Hence, a flatus-free rectum is not indica-tive of a proximal luminal blockage.

MEASUREMENT OF DILATATION

Many radiologists and surgeons believe that a small-bowel loop measuring greater than 2.5 cm in diameteris indicative of distention and is often associated withluminal blockage. However, the reliance on this mea-surement is based on a fallacy. Remember that a con-ventional x-ray emanates from a point source, and theresultant image is displayed on the flat surface of animaging plate placed behind the patient. Thus, magni-fication is always a factor. Also, one cannot determineon a single supine view whether a loop of bowel thatmay be distended is in fact close to the film, andtherefore far from the point source, or if it is moreventrally situated. Moreover, one also does not knowhow thick the patient is to determine how far anteri-orly the ventral bowel loop may be situated. For ex-ample, in an obese patient measuring 20 inches (51cm) from front to back at the umbilicus, an anteriorlysituated bowel loop will be magnified by at least a fac-

Figure 6. In the appropriate clinical context, the diagnosis of intestinal obstruction can often be made equally well with conventionalradiography and CT. (a) Supine conventional radiograph shows dilated small bowel and nondilated ascending colon. (b) TransverseCT image shows mostly fluid-filled distended small-bowel loops and nondilated ascending colon.

Co

nven

tion

al Radio

grap

hs o

f the A

cute A

bd

om

en

139

tor of two and will appear to be dilated even when itis actually of normal caliber (14).

TUBE IN PLACE: RELIEF OF DILATATION

The determination of bowel obstruction on conven-tional radiographs depends on the differential disten-tion of gas in loops proximal to the point of occlusionand a lack of distention in loops distal to it. Thus, thecontrast agent used for the determination of bowelobstruction at conventional radiography is primarilyswallowed air. An effective treatment in many cases ofbowel obstruction before or in lieu of surgery is toplace a nasogastric tube into the stomach or more dis-tally into the proximal duodenum. The function ofthe tube is to remove accumulations of intraluminalgas. With such removal, the patient may feel better,and the resulting radiograph could show a decrease inthe distention of the bowel proximal to the obstruc-tion. However, the relief of obstruction is occasionedby the passage of gas through the anus. It is not deter-mined by a decrease in the amount of gas proximal tothe occlusion even if a tube is in place and even if thetreatment makes the patient feel better. Thus, it is fairto say that the dilatation has been ameliorated but notthat the obstruction has necessarily been resolvedwhen a tube is in place.

EARLY OBSTRUCTION

When the conventional radiograph reveals gas in anondilated loop distal to an obstruction in associa-tion with proximal distended loops, the temptationoften is to claim that the luminal occlusion is early orrecent. One cannot make this statement with any va-lidity because on one radiograph, the temporal course

of obstruction cannot be determined or even sug-gested. It is just as conceivable that what is being ob-served is persistent intermittent obstruction ratherthan an obstruction seen early in its course (Fig 7).

PARTIAL OR INTERMITTENT OBSTRUCTION

The term partial obstruction is one long favored bysurgeons. However, the surgical literature provides nodefinition of partial obstruction in the setting of anacute clinical situation. The term presupposes thatsomehow the lumen is narrowed but not completelyblocked, thereby not completely hindering the pas-sage of gas distally. This notion may be rational inconception, but it has no empiric verification. It is en-tirely likely that what is deemed “partial obstruction”may be a complete but intermittent obstructioncaused by the reversible sharp bending of the bowelnear the point of occlusion as a result of peristalsis.However, although the dynamic of complete but in-termittent obstruction seems plausible, it, too, has yetto be proved. Thus, because partial, incomplete, or in-termittent blockages are not amenable to confirma-tion, these terms should not be used to further charac-terize an intestinal occlusion. It is enough to say thatan obstruction exists.

POINT OF OBSTRUCTION

With conventional radiographs, one cannot determinethe actual point of luminal occlusion. Almost always,many liquid-filled loops are interposed between thesite of blockage and the most distal gas-filled loopthat can be detected radiographically. Consequently,a single loop of dilated jejunum may be associatedwith a jejunal obstruction, but it is also possible thatthe dilated jejunal loop is a concomitant of an ilealor colonic obstruction. Hence, except for rare cases inwhich an additional pathognomonic finding ispresent, determination of the actual site of bowel oc-clusion on a conventional radiograph should not beattempted. Here, too, it is enough to recognize thepresence of bowel occlusion (Fig 8).

SENSITIVITY OF THE CONVENTIONALABDOMINAL RADIOGRAPH AND EVALUATIONOF OBSTRUCTION

Clearly, CT is more versatile than conventional radi-ography in assessing some of the important character-istics of a bowel obstruction. CT can be used to deter-mine the presence and, at times, to find the site andthe cause of obstruction; whereas, some rare excep-tions withstanding, conventional radiography canonly distinguish the presence of an obstruction.

Nonetheless, the results of two studies have showna conventional radiographic sensitivity of 66% in one

Figure 7. Early obstruction versus intermittent obstruction.Radiograph shows dilated small bowel and nondistention of co-lon 1 day after admission. From these findings, it was thoughtthat the bowel recently became obstructed. A more completehistory revealed that the obstructive symptoms had persistedover several days.

Bak

er

140

study and 68.8% in the other (15,16). The latter studyalso assessed contemporaneous CT. The findings re-vealed that CT had a sensitivity of only 64% for therecognition of bowel obstruction (16). Of course, CTis more specific, but its overall accuracy was similar tothat of conventional radiography. Therefore, even inthis present era of abundant and steadily increasinguse of multi–detector row CT, which will soon in-clude standard coronal reformation in addition tosagittal and transverse views, the conventional radio-graph stands up well with respect to the crucial issueof the recognition of the presence of bowel blockage.

PROGRESSION OF PATIENTS THROUGHEMERGENCY DEPARTMENT EVALUATION

The customary progression of a patient with acute ab-domen, as practiced in many medical centers, beginswith conventional radiographs of the abdomen. Theimages may be assessed separately, yet often they areobtained but not immediately interpreted. The patientis then transported from the standard radiographyroom to a separate suite in which a CT examination isperformed, sometimes after contrast agents are admin-istered orally and intravenously. As part of the CT ex-amination, a digital view of the abdomen is obtainedin most instances prior to the generation of CT sec-tions. Several issues in this ritual should be assessed toprovide optimal care at reasonable dose accumulation.

In actual practice, the CT scout view is superior tostandard radiographs of the abdomen for several rea-sons. Some have complained that this examinationdoes not reveal subtle findings of calcification or lu-cency. However, in our studies comparing the supineradiograph to CT images for free-air detection, wehave used the digital scout view as the plain radio-graphic examination of choice because the CT exami-nation could not be delayed, nor could the patient bemoved. We have found that the conventional radio-graph equals CT images for the recognition of pneu-moperitoneum. Furthermore, the CT scout view en-compasses the entire abdomen, overcoming the prob-lem intrinsic with abdominal radiographs obtainedwith a 40-inch source-to-film distance. Therefore, webelieve that the stand-alone abdominal series shouldbe bypassed and that the CT scout view is an integralpart of the radiographic work-up.

For intestinal obstruction, the next most importantview obtained after a supine radiograph was shownby Bryk in the 1970s to be successive supine radio-graphs (17). Most of the time, we can make the diag-nosis of the presence of obstruction, if it was in doubtin the first radiograph, by looking at the CT scoutview. It is often more revealing than the initial radio-graph because in the interval between the two, the pa-tient has had time to become more anxious and thushas had the opportunity to swallow air to a greater ex-

tent, enabling the successive supine radiograph to bemore revealing.

The problem with this scenario is that reimburse-ment is provided for the supine radiographic studiesobtained in a standard radiography room and also forthe CT examination that encompasses the digitalscout view. If the digital scout view is obtained as partof the CT examination and yet considered as a sepa-rate examination, it will not be reimbursed. Nonethe-less, that is precisely the stepwise way that we proceedin our imaging evaluation: We assess the digital scoutview before the cross-sectional views are inspected.

Consequently, we propose that the digital scoutview be looked at as a separate examination for billingpurposes, if it was done alone and if it determined thepresence of obstruction or perforation. If the digitalscout view was part of a subsequent CT study, itshould not be billed separately but encompassed inthe charge for the entire study. If separate reimburse-ment is established for the digital scout view, thischange in payment may give impetus to reduce therate of CT utilization. However, there is probably littleenthusiasm for such an alteration of professionalcompensation today because it is counterproductiveto the accustomed means of generation of the incomeof radiologists.

When the debate about CT use becomes more in-tense, especially when focused studies in young pa-tients reveal a relationship between exposure and later

Figure 8. Dilated gas-filled colon to the sigmoid. Radiographshows that in actuality, the obstruction was in the rectum,more distal than the distribution of gas shadows would seemto indicate.

Co

nven

tion

al Radio

grap

hs o

f the A

cute A

bd

om

en

141

malignancy, radiologists will need to make behavioralchanges in acknowledgment of the implication ofdose accumulation with CT. We have recently wit-nessed an effort by manufacturers to reduce x-ray ex-posure in pediatric patients in response to the public-ity recently directed at the dose from CT machines.However, no real change has occurred in the activityof radiologists or in initiatives with respect to the re-striction of unindicated studies. The next phase of thediscussion about dose will inevitably be disquietingto those who favor the status quo. A new approach toreimbursement for the CT scout view (the modernconventional radiograph) may be a consideration inthe upcoming debate.

References1. Baker SR, Hirschorn D, Jacobs AF. The recognition of

miniscule intra abdominal gas bubbles: comparison of su-pine radiographs, CT scouts and CT sections (abstr). Radi-ology 1999; 213(P):252–253.

2. Miller RE. The technical approach to the acute abdomen.Semin Roentgenol 1973; 8:267–279.

3. Miller RE, Becker GJ, Slabaugh RA. Detection of pneumo-peritoneum: optimum body position and respiratory phase.AJR Am J Roentgenol 1980; 135:487–490.

4. Woodring JH, Heiser MJ. Detection of pneumoperitoneumon chest radiographs: comparison of upright lateral poste-rior-anterior projections. AJR Am J Roentgenol 1995; 165:45–47.

5. Stapakis JC, Thickman D. Diagnosis of pneumoperitoneum:abdominal CT vs upright chest film. J Comput AssistTomogr 1992; 16:713–716.

6. Nirapathpongporn S, Osatavanichvong K, Udompanich O,Pakanan P. Pneumoperitoneum detected by ultrasound.Radiology 1984; 150:831–832.

7. Levine MS, Scheiner JD, Rubesin SE, Laufer I, Herlinger H.Diagnosis of pneumoperitoneum on supine abdominal ra-diographs. AJR Am J Roentgenol 1991; 156:731–735.

8. Baker SR. Imaging of pneumoperitoneum. Abdom Imaging1996; 21:413–414.

9. Cho KC, Baker SR. Extraluminal air: diagnosis and signifi-cance. Radiol Clin North Am 1994; 32:829–844.

10. Keeffe EJ, Gagliarki RA, Pfister RC. The roentgenographicevaluation of ascites. Am J Roentgenol Radium Ther NuclMed 1967; 101:388–396.

11. Frimann-Dahl J. Roentgen examinations in acute abdominaldiseases. 2nd ed. Springfield, Ill: Thomas, 1960.

12. Bryk D. Functional evaluation of small bowel obstruction bysuccessive abdominal roentgenograms. Am J RoentgenolRadium Ther Nucl Med 1972; 116:262–275.

13. Mindelzun RE, McCort JJ. Questions and answers. AJR AmJ Roentgenol 1996; 166:716–718.

14. Baker SR. Unenhanced helical CT versus plain abdominalradiography: a dissenting opinion. Radiology 1997; 205:45–47.

15. Wittenberg J. The diagnosis of colonic obstruction on plainabdominal radiographs: start with the cecum, leave the rec-tum to last. AJR Am J Roentgenol 1993; 161:443–444.

16. Maglinte DD, Reyes BL, Harmon BH, et al. Reliability androle of plain film radiography and CT in the diagnosis ofsmall-bowel obstruction. AJR Am J Roentgenol 1996; 167:1451–1455.

17. Maglinte DD, Balthazar EJ, Kelvin FM, et al. The role of ra-diology in the diagnosis of small-bowel obstruction. AJR AmJ Roentgenol 1997; 168:1171–1180.

NO

TES

143

Imaging of CervicalSpine Trauma1

The objective of this chapter is to introduce an evidence-based approach to imaging ofthe cervical spine in blunt trauma. In particular, I will focus on the role of computedtomography (CT) and on the reasons that we elect to use either CT or radiography indifferent patients. I will also discuss injury patterns in the cervical spine, dividing thecervical spine into upper and lower components and identifying several problem areaswhere there are challenges to the interpretation of CT or radiographic studies.

SELECTION OF APPROPRIATE IMAGING APPROACH

Trauma victims who come to the emergency department need to have their cervicalspines evaluated to exclude the presence of an unstable fracture that could progress toneurologic compromise. In some cases, tools such as the National Emergency X-Radiog-raphy Utilization Study (NEXUS) (1) or the Canadian cervical spine rule (2) can beused to identify patients who do not require imaging but instead might have their cer-vical spines cleared purely on clinical grounds. However, most patients will requireimaging with either CT or radiography.

For decades, the standard method of clearing the cervical spine has been with radi-ography. A series of radiographs consisting of anteroposterior (including open-mouthodontoid) and lateral (including swimmer) views that show the spine from the baseof the skull to the junction of the seventh cervical vertebra with the first thoracic verte-bra can be used to exclude the presence of cervical spine fracture. Radiography hasmany advantages. It is relatively inexpensive, is available essentially everywhere thatthere is an emergency department, and has been around sufficiently long that there isinterpretation expertise at nearly all centers. However, radiography struggles in pa-tients who are at the highest risk of fracture. In patients who have sustained majortrauma, the presence of spine immobilization backboards, other injuries that preventoptimal positioning (including upper extremity fractures and head injury), and life-support apparatus (such as endotracheal tubes) makes obtaining adequate evaluationof the cervical spine with radiography challenging (3). In addition, patients with ma-jor trauma may be uncooperative because of hypoxia, intoxication, or head injury. Inpatients with major trauma, radiography, instead of requiring 10 minutes to com-plete, may require an hour and may still lead to incomplete or inadequate imagingstudies (4).

C. Craig Blackmore, MD, MPH

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 143–149.

1From the Department of Radiology, Harborview Medical Center, Harborview Injury Prevention and Research Center,University of Washington, Box 359960, 325 Ninth Ave, Seattle, WA 98104 (e-mail: craige@u.washington.edu).

Bla

ckm

ore

144

In the mid-1990s, CT was promoted as an alterna-tive to radiography for clearing the cervical spine intrauma patients (5). CT offers several advantages. It ismore sensitive for fracture (6–8), is more specific inpatients with major trauma, and is fast, particularly insubjects who are already positioned on the CT gantryfor head CT (5). Although the amount of collective ex-perience with CT is less than that with radiography,most fractures fortunately are readily apparent on CT.In addition, the use of coronal and sagittal reforma-tions, particularly after multi–detector row CT scan-ning, can enable adequate evaluation of the cervicalspine, even in the presence of other injuries, life-sup-port devices, and preexisting conditions such asosteopenia and kyphosis.

Obviously, there is a trade-off between CT and radi-ography. CT is more sensitive, more specific, and fasterbut does have a higher cost, especially when one con-siders the direct cost of the imaging involved. However,CT becomes cost-effective if subjects are identified whoare at high (>4%) probability of fracture (9). The cost-effectiveness of CT is due to the frequency of inad-equate radiographs, the extreme cost of a missed frac-ture (even though only a small percentage of patientswill develop neurologic deficit, such as paralysis), andthe higher cost of radiography in high-risk subjects thatis caused by the requirement for multiple repeat views.It is possible to identify subjects who are at high risk byusing the Harborview CT screening criteria (10), a vali-dated clinical prediction rule that is based on the pres-ence of focal neurologic deficit, severe head injury, or ahigh-energy trauma mechanism. Thus, CT is accurate,rapid, and effective as a screening strategy for the cervi-cal spine in trauma patients and becomes cost-effectivewhen criteria, such as the Harborview cervical spinescreening criteria, are used to identify appropriatehigh-probability patients. Radiography remains opti-mal (a) for subjects at low risk or (b) if rapid (helical)CT is not available (11).

UPPER CERVICAL SPINE

The upper portion of the cervical spine, particularlythe craniocervical junction, is one of the most fre-quently injured areas of the cervical spine. Further,this region is difficult to depict with radiography.With the introduction of CT screening, injuries to thisregion are being diagnosed more frequently. There-fore, radiologists need to be aware of the injury pat-terns and osseous and ligamentous anatomic struc-tures. Many of the fractures of the craniocervical junc-tion are avulsion-type injuries, emphasizing theimportance of the ligamentous structures.

The major ligaments of the craniocervical junctioninclude the occipital condylar articulation capsularligaments, the apical ligament from basion to dens,the alar ligaments from occipital condyles to dens, the

cruciate ligaments, including both the transverse andhorizontal components, and the tectorial membrane,representing the extension of the posterior longitudi-nal ligament cephalad.

Occipital condyle fractures are classically divided intothree types. The first type is a burst fracture. These burstfractures result from axial load with compression of thecondyle but generally involve intact atlanto-occipitalcapsular ligaments and are therefore stable. Type 2occipital fractures represent extension of an occipitalbone fracture into the condyle. Type 3 occipital con-dyle fractures are the most common (Fig 1). These areavulsions from the alar ligament (12). Stability of oc-cipital condyle fractures remains a controversial topic(13). There are no good data on the natural history ofthese injuries to define which are stable. This problemis exacerbated by the fact that until the past severalyears, occipital condyle fractures were thought to berare. With CT screening, such fractures have becomeone of the more commonly identified injuries of thecraniocervical region. Current definitions of stabilityare related to the extent of displacement of avulsion in-juries, with displacement of less than 5 mm consideredstable. In addition, any evidence of disruption of thecapsular ligaments is a criterion for instability, and ingeneral, bilateral injuries are considered unstable. Treat-ment for these injuries remains controversial. Instabil-ity at the atlanto-occipital joint may be subtle. How-ever, sagittal reformations and coronal reformations ofthe CT scan data provide a side-to-side comparison thatcan facilitate identification of subtle atlanto-occipitalinjury. Magnetic resonance (MR) imaging may also beindicated to evaluate these injuries.

Figure 1.(a) Axial CTimage and(b) coronalreformationshow fracture(arrow) of leftoccipital con-dyle fromavulsion ofalar ligament.

Cervical Sp

ine Trau

ma

145

Fractures of the first cervical vertebra ring (Fig 2) aregenerally related to axial loading and commonly occurin association with other injuries, including those ofthe occipital condyle and the axis. Neurologic compro-mise is relatively infrequent with fractures of the cervi-cal vertebra C1 ring, presumably because the axial com-pression mechanism results in a burst configurationwith expansion of the spinal canal. Burst fractures aredefined as stable if the transverse component of the cru-ciate ligaments, which is also referred to as the trans-verse atlanto-axial ligament, is intact. Integrity of thetransverse ligament is inferred from the displacementof the lateral masses at cervical vertebra C1. Greaterthan 7 mm of displacement or an avulsion fracture ofthe cervical vertebra C1 tubercle is considered ligamen-tous injury. Displacement is assessed by evaluating theoverlap of the lateral masses of the cervical vertebra C1upon the lateral masses at cervical vertebra C2.

Fractures of the atlas are divided into two majorgroups: (a) fractures isolated to the dens and (b) frac-

tures involving the posterior elements. Dens fracturesare frequent, although the mechanism of injury iscomplex and not well understood. Dens fractures arecommonly broken down into three types. Type 1 inju-ries are avulsion fractures of the tip of the dens, eitherfrom alar or apical ligament avulsion. These injuriestend to be stable, although like occipital condyle frac-tures, bilaterality connotes instability. There is an as-sociation between type 1 dens fractures and cervicalvertebra C1 ring fractures. Type 2 dens fractures arefractures across the base of the dens (Fig 3) and areparticularly common and problematic in the elderly.Type 3 dens fractures involve some portion of thebody of the dens and may include the articular facetbut do not involve the posterior elements (14).

Avulsion fractures can also occur from the attach-ment of the anterior longitudinal ligaments on theanterior inferior corner of cervical vertebra C2. Thesehyperextension teardrop injuries may be subtle, andthey are generally stable. It is important, however, todifferentiate the stable anterior longitudinal ligamentavulsion from the potentially unstable anulus fibro-sus avulsion with disk dislocation. The differentiatingfactor between these two is that the stable anteriorlongitudinal ligament hyperextension teardrop avul-sion fracture will have normal alignment at the C2–3disk space. Any evidence of abnormal alignment atthis disk space should raise suspicion for C2–3 dislo-cation and should be considered unstable. Detectionof cervical vertebra C2 hyperextension teardrop frac-tures is a potential pitfall for CT scanning. Whenthese injuries are subtle and nondisplaced, they mayoccasionally be difficult to depict with CT scanning.

Fractures involving the posterior elements of cervi-cal vertebra C2 generally fall into the category of theC2 traumatic spondylolisthesis, also called the hang-man fracture (Fig 4). These fractures tend to occurfrom a hyperextension mechanism, although this, too,is controversial. In general, the canal is enlarged withhangman fractures, and neurologic compromise is un-common. Sometimes the fracture will involve a por-tion of the posterior vertebral body. In these varianthangman fractures, the involved component of theposterior vertebral body may undergo retropulsionand may contribute to cord injury.

LOWER CERVICAL SPINE

The anatomic structures and injury patterns from cervi-cal vertebra C3 to cervical vertebra C7 are relatively ste-reotypical. Anatomically, the lower portion of the cer-vical spine can be divided into anterior and posteriorcolumns. Biomechanical stability is defined as involve-ment of all of the osseous elements of one column plusone element of the other column. Instability may alsobe defined by (a) olisthesis of 3.5 mm or more or(b) focal kyphosis at a single level of 11° or more (15).

Figure 3.Sagittal CTreformationdemon-strates type2 fracture ofdens, withnearly 100%posterior dis-placement ofdens.

Figure 2. Axial CT image shows burst fracture (arrows) of cer-vical vertebra C1. Stability and integrity of transverse ligamentare inferred from absence of widening of lateral masses with re-spect to dens.

Bla

ckm

ore

146

There is no universal classification system for thelower portion of the cervical spine. Several systemshave been proposed, which are based both on the di-rection of force and the resultant injury patterns. Thesystem that will be used in this discussion is based onmanagement practices at our institution (16). We willdivide injuries of the lower portion of the cervical spineinto minor stable fractures, facet fractures with normalalignment, facet dislocations, burst fractures, and exten-sion injuries. Minor stable fractures include spinousprocess fractures, extraforaminal transverse process frac-tures, and fractures isolated to the anterior column. An-terior column fractures (Fig 5) generally involve onlythe end plate, with little or no anterior vertebral bodyheight loss, and are generally treated conservatively witha cervical collar. Spinous process fractures, particularlyat cervical vertebrae C6 and C7, may occur in isolationfrom forces applied by the interspinous or nuchal liga-ments. These stable injuries are generally treated con-servatively. Management of fractures of the transverseprocess (Fig 6) is more controversial, although the frac-ture itself, if isolated to the transverse process, will bebiomechanically stable; however, the proximity of thevertebral arteries to these injuries raises the specter of

vertebral artery injury and heightens the risk of verte-bral basilar system stroke. Some investigators advocatevascular imaging in all transverse process fractures, anapproach that remains controversial (17–19).

Fractures through the articular facets occur as a conse-quence of lateral bending or rotation. These fracturesmay be isolated but with disruption of the capsularligaments may still be potentially unstable and are of-ten treated with halo external-fixation devices. Facetfractures are often difficult to depict with radiographyand represent an area where CT has higher sensitivity.

Dislocations at the facet joint, by definition, involvedisruption of the capsular ligaments and therefore arepotentially unstable. Dislocations include “jumped”facets, where the inferior articular surface of the uppervertebral body is dislocated anterior to the superior ar-ticular facet of the lower vertebral body. Such disloca-tions may or may not be accompanied by fracture.“Perched” facets occur when the apex of the articularfacets of adjacent vertebral levels are in apposition. In-juries with such large displacements of the articularfacet usually lead to olisthesis. In general, unilateralfacet dislocation (Fig 7) will lead to canal compromiseof approximately 25% of the diameter, and bilateral

Figure 4. (a) Axial and(b) sagittal CT images showtraumatic spondylolisthesis ofcervical vertebra C2. Involve-ment of posterior margin of ver-tebral body (arrow) increasesprobability of neurologic compro-mise.

Figure 5. (a) Axial and (b) sag-ittal CT images show anteriorend-plate fracture (arrow) withpreserved alignment. Usually,these injuries are stable and aremanaged conservatively.

Cervical Sp

ine Trau

ma

147

facet dislocation will lead to canal compromise of 50%(Fig 8). Because of the potential for cord ischemia insuch high degrees of canal compromise, immediate re-duction is advocated. In general, we do not perform de-finitive imaging (eg, CT, MR imaging) in these patientsuntil fluoroscopy-guided reduction has been per-formed by the physicians of our spine service.

Burst fractures are a consequence of axial load, typi-cally with an element of flexion, and these fracturesresult in crushing and expansion of the vertebralbody (Fig 9). Canal compromise often occurs, andthere may be an additional distraction component tothe injury in the posterior elements. Often burst frac-tures will result in a triangular bone fragment fromthe anterior vertebral body. This fragment bears someresemblance to a teardrop; hence, the common termflexion teardrop fracture is applied to these cases. Theprobability of neurologic deficit with such injuries ishigh. MR imaging may be indicated to evaluate theneural elements and to evaluate for associated dis-traction injuries of the posterior ligaments.

Finally, extension injuries can occur in the cervicalspine. As in other portions of the spine, extension inju-ries are more common and more severe in subjectswith abnormal fusion of the spine. The classic exampleis ankylosing spondylitis, in which the syndesmophytesbridging the disk space and the ankylosis of the facetarticulations lead to a rigid spine. Failure of energydiffusion across multiple vertebral levels leads to frac-ture at lower energy in such subjects. Hyperextensionfractures tend to be severely displaced, and neurologiccompromise is common. A subcategory of extensioninjuries includes those that occur without fracture.These injuries are the so-called SCIWORA (SpinalCord Injury Without Radiologic Abnormality) inju-ries. These injuries tend to occur in two groups of pa-tients. The first group is children, in whom hyper-mobility of the cervical spine related to immaturitycan allow cord injury without fracture or ligamentousdisruption. The second category is patients with spinalstenosis, as in ossification of the posterior longitudinalligament. In these patients, the narrowing of the canal

Figure 6. (a) Lateral radio-graph shows soft-tissue swelling(arrows) that is the only radio-graphic evidence for fracture.(b) Axial CT image shows subtlefracture through left transverseprocess at cervical vertebra C2.Although stable and not requiringspecific treatment, fracturesthrough transverse foramenraise suspicion for vertebral ar-tery injury.

Figure 7. (a) Axial and (b) sag-ittal CT reformations show uni-lateral facet dislocation (perch)(arrow).

Bla

ckm

ore

148

is exacerbated by physiologic motion, and cord injurycan result.

In conclusion, the intent of this chapter was two-fold. The first purpose was to introduce an evidence-based imaging approach to the cervical spine andtrauma patients. CT is the preferred imaging modalityin subjects at high risk of injury because of the highersensitivity and specificity of CT. However, radiographyis still preferred in low-risk subjects because it is morecost-effective and because the radiation dose is lower.The second purpose was to discuss injury patterns inthe upper and lower portions of the cervical spine,calling attention to challenging areas in the interpreta-tion of both CT and radiographic studies.

Figure 9.(a) Axial and(b) sagittalCT imagesshow burstfracture ofcervical ver-tebra C6,with retropul-sion of os-seous frag-ments andcanal com-promise.

Figure 8.(a) Axial CTimage showsbilateral facetdislocationwith olisthe-sis. (b) Sagit-tal reforma-tion demon-strates canalcompromise.Reduction isusually per-formed priorto CT. How-ever, inter-facet os-seous frag-ments canprevent ad-equateclosed reduc-tion, as inthis case.

References1. Hoffman J, Mower W, Wolfson A, Todd K, Zucker M. Validity

of a set of clinical criteria to rule out injury to the cervicalspine in patients with blunt trauma. N Engl J Med 2000; 343:94–99.

2. Stiell I, Wells G, Vandemheen K, et al. The CanadianC-spine rule for radiography in alert and stable trauma pa-tients. JAMA 2001; 286:1841–1848.

3. Blackmore CC, Deyo RA. Specificity of cervical spine radi-ography: importance of clinical scenario. Emerg Radiol1997; 4:283–286.

4. Blackmore CC, Zelman WN, Glick ND. Resource costanalysis of cervical spine trauma radiography. Radiology2001; 220:581–587.

5. Nunez DB, Ahmad AA, Coin CG, et al. Clearing the cervicalspine in multiple trauma victims: a time-effective protocol

Cervical Sp

ine Trau

ma

149

using helical computed tomography. Emerg Radiol 1994;1:273–278.

6. Nunez DB, Quencer RM. The role of helical CT in the as-sessment of cervical spine injuries. AJR Am J Roentgenol1998; 171:951–957.

7. Ptak T, Kihiczak D, Lawrason J, et al. Screening for cervicalspine trauma with helical CT: experience with 676 cases.Emerg Radiol 2001; 8:315–319.

8. Hanson JA, Blackmore CC, Mann FA, Wilson AJ. Cervicalspine injury: accuracy of helical CT as a screening tech-nique. Emerg Radiol 2000; 7:31–35.

9. Blackmore CC, Ramsey SD, Mann FA, Deyo RA. Cervicalspine screening with CT in trauma patients: a cost-effec-tiveness analysis. Radiology 1999; 212:117–125.

10. Hanson J, Blackmore CC, Mann FA, Wilson AJ. Cervicalspine injury: a clinical decision rule to identify high-risk pa-tients for helical CT screening. AJR Am J Roentgenol 2000;174:713–717.

11. Blackmore CC, Mann FA, Wilson AJ. Helical CT in the pri-mary trauma evaluation of the cervical spine: an evidence-based approach. Skeletal Radiol 2000; 29:632–639.

12. Anderson PA, Montesano PX. Morphology and treatment ofoccipital condyle fractures. Spine 1988; 13:731–736.

13. Hanson J, Deliganis A, Baxter A, et al. Radiologic and clini-cal spectrum of occipital condyle fractures: retrospective re-view of 107 consecutive fractures in 95 patients. AJR Am JRoentgenol 2002; 178:1261–1268.

14. Anderson LD, D’Alonzo RT. Fractures of the odontoid pro-cess of the axis. J Bone Joint Surg Am 1974; 56:1663–1674.

15. White AA, Southwick WO, Panjabi MM. Clinical instability inthe lower cervical spine. Spine 1976; 1:15–27.

16. Anderson P. Spine. In: Hansen S, Swiontkowski M, eds. Or-thopaedic trauma protocols. New York, NY: Raven, 1993;211–216.

17. Biffl WL, Moore EE, Elliott JP, et al. The devastating poten-tial of blunt vertebral arterial injuries. Ann Surg 2000; 231:672–681.

18. Biffl WL, Ray CE Jr, Moore EE, et al. Treatment-related out-comes from blunt cerebrovascular injuries: importance ofroutine follow-up arteriography. Ann Surg 2002; 235:699–706; discussion 706–707.

19. Azuaje R, Jacobson L, Glover J, et al. Reliability of physicalexamination as a predictor of vascular injury after penetrat-ing neck trauma. Am Surg 2003; 69:804–807.

NO

TES

151

Imaging Spine Traumain the Elderly1

National demographics show an increasingly larger number of older individuals inindustrialized societies. In general, older individuals may be considered as “elderly”(aged ≥65 years) on physiologic and epidemiologic bases that reflect observed differ-ences in functional decline, trauma mortality rates, and osteoporosis (1–4). Duringthe past century, the number of persons in the United States older than age 65 in-creased 10-fold, and rapid growth is forecast with the baby-boom generation reachingretirement age starting in the year 2011 (5).

In general, spine fractures and resultant spinal cord injuries are an important sourceof morbidity and mortality. In 1993, spinal cord injuries occurred at a rate of approxi-mately 30 per 1 million person-years and cost society an estimated $3.4 billion (6,7).The elderly experience these injuries disproportionately. The results of a prospective sur-vey in Taiwan revealed that spinal cord injuries occurred at a rate of approximately 48per 1 million person-years in patients older than 65 years, compared to 19 per 1 millionperson-years in patients younger than 65 years (8). Moreover, injury patterns in elderlypatients may differ from those in younger patients because of differences in bone densityand injury mechanisms and because of the presence of degenerative changes (9).

The purpose of our review is to (a) describe individual patient and mechanism fac-tors that may affect spine injury in the elderly population, (b) discuss the influence ofsenescent changes on biomechanics and fracture patterns in the elderly spine, (c) illus-trate common patterns of spine injury in elderly patients, and (d) analyze current im-aging techniques for evaluation of spine injury with respect to elderly patients.

INDIVIDUAL PATIENT AND MECHANISM FACTORS

In the United States, almost 3 million individuals are admitted to the hospital fortrauma annually (3). Among these admitted trauma patients, approximately 10.000cervical spine fractures (incidence, approximately 15–30 per 1 million individuals peryear) and 4000 thoracolumbar spine fractures (incidence, approximately 10–20 per 1million individuals per year) are diagnosed. Within these groups of patients, approxi-mately one-third and one-quarter, respectively, sustain neurologic injury (10). Thus,spinal cord injury is relatively rare.

With the exception of the elderly, traumatic spinal cord injury results from high-en-ergy mechanisms (eg, motor vehicle accidents, falls from heights greater than 3–4 m).

Friedrich M. Lomoschitz, MD, C. Craig Blackmore, MD, MPH,and Frederick A. Mann, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 151–158.

1From the Department of Radiology, Vienna Medical School, University of Vienna, Waehringer Guertel 18-20, A-1090Vienna, Austria (F.M.L.); and the Department of Radiology, Harborview Medical Center, Seattle, Wash (C.C.B., F.A.M.)(e-mail: friedrich.lomoschitz@univie.ac.at).

Lom

osc

hit

z et

al

152

Among the elderly, lower-energy impacts, such as fallsfrom seated or standing heights, are a common causeof clinically unstable spine injuries (11). Cervical spineinjuries in elderly patients tend to involve more thanone level with consistent clinical instability, and theseinjuries commonly occur at the atlantoaxial complex(12). Compared with younger individuals, the elderlyare 2–12 times more likely to be injured in domesticfalls (13–15) and are 2–4 times more likely to sustaininjuries of the upper portion of the cervical spine (cer-vical vertebrae C1 and C2) (13,14,16). Fractures oc-curring at the atlantoaxial complex in elderly patientsalmost always involve cervical vertebra C2, and anisolated fracture of cervical vertebra C1 is rare (17).Moreover, elderly patients also are more likely to havefractures overlooked at initial diagnosis (15%–40% vs4% in younger individuals) (15,16).

In combination with the thoracic cage, the thoracicspine is inherently stable, and traumatic thoracicspine fractures are less common than fractures of thecervical and lumbar regions, which is also true for theelderly population. In the general population, aboutevery second patient with a thoracic spine fracture hasaccompanying neurologic findings. This rate is due tothe fact that injuries that result in thoracic spine frac-ture are usually caused by high-energy trauma. As a re-sult, particularly in the upper portion of the thoracicspine, fracture-dislocations are more common thancompression fractures and burst fractures. Moreover,the size of the thoracic spinal cord is large relative tothe small spinal canal, and the size of the spinal canalmay even be decreased by accompanying degenerativechanges in elderly patients (18). Falls are the majorcause of fractures in the lower portion of the thoracicspine and the lumbar spine, with burst fractures beingthe most frequent type of fracture encountered. In pa-tients who are injured in falling accidents, age seemsto have no influence on the type and location of tho-racolumbar spine fractures (19).

INFLUENCE OF SENESCENT CHANGES ONBIOMECHANICS AND FRACTURE PATTERNS

Age-related comorbidities, such as concurrent medicalillness and dementia, may distract from an accurateand reliable clinical evaluation. Spondylosis or os-teoporosis distorts vertebral anatomic structures andmay render normal standard paradigms less useful indetecting injury, may create structures outside thestandard paradigms usually used in pattern recogni-tion, and may obscure radiographic signs of trauma.

All of the fractures that occur in the elderly also arefound in younger individuals. However, there is astriking change in the frequency of some specific frac-tures in the elderly. In older individuals, among bothmen and women, there is a dramatic increase in thefrequency of fractures of the atlantoaxial complex (ie,cervical vertebrae C1 and C2).

Osteoporosis and senescent degenerative disordersappear to predict the apparently lower force thresholdfor fractures. The frequency of osteopenia increaseswith age, especially for postmenopausal women. Inthe presence of osteoporosis, bone loss is global andequally affects the whole spine. However, normal se-nescent cortical and trabecular bone losses are quanti-tatively less in the cervical spine compared with thethoracolumbar spine (20).

Craniocervical Junction

The biomechanical response of the elderly spine toblunt trauma is different from that of the spine inyounger patients. In the cervical spine, senescent degen-erative changes tend to occur in the mid and lower por-tions of the cervical spine, allowing a relatively greaterdegree of mobility (a) to the craniocervical junction, in-cluding the motion segments of the occipital condylesthrough cervical vertebra C2, and (b) particularly to theatlantoaxial complex, which is where fractures most of-ten occur in the elderly (9).

Figure 1. "High" (Anderson-D’Alonzo type II) odontoid fractureand incomplete spinal cord injury in an 80-year-old man injuredas a restrained driver in high-speed motor vehicle accident, withendotracheal and orogastric intubation performed at scene ofaccident. Lateral radiograph shows type II dens fracture (arrow),with posterior angulation and displacement of dens (outlined by■). In lower portion of cervical spine, severe degenerativechanges (chronic degenerative disk disease and diffuse idio-pathic skeletal hyperostosis) are present.

Spin

e Traum

a in th

e Elderly

153

Degenerative osseous changes are known to influ-ence the site of cervical spine injuries and are assumedto be present in almost all patients older than 65 years.Normally, in the younger individual, the most mobilecervical motion segments are cervical vertebrae C4through C7. Not surprisingly, most cervical fractures inyounger patients occur at these levels (8). With degen-erative changes, these same segments become less mo-bile, and the motion segment of cervical vertebrae C1and C2 becomes the most mobile portion (9). Thehigher incidence of injury to the upper portion of thecervical spine in the elderly population may be due tothe stiffening effect of aging on the vertebral column.

Thoracolumbar Junction

The transition from the thoracic spine to the upperportion of the lumbar spine is referred to as the thora-columbar junction (ie, thoracic vertebra T10 throughlumbar vertebra L2). This transition zone, in general,is exposed to injury because of several factors, includ-ing the absence of a protective rib cage, a change inthe alignment from kyphosis to lordosis, and a changein the orientation of the facet joints from a coronalorientation in the thoracic spine to a more oblique tosagittal orientation in the lumbar spine. Senescentchanges, including osteoporosis, may lower the forcethreshold for fractures in this region in the elderlypopulation. However, also in the general population,wedge-compression fractures are the most commoninjury pattern in the thoracic and lumbar spine, usu-ally occurring near the thoracolumbar junction.

COMMON PATTERNS OF SPINE INJURYIN THE ELDERLY

The leading cause of cervical injury in older patients islow-energy trauma, mainly as a result of falls fromstanding or seated height (8,11,14). The elderly have a

high frequency of injury of the upper portion of the cer-vical spine, particularly fractures involving the atlanto-axial complex (16,21,22), which may be expected onthe basis of simple biomechanics.

Cervicocranial Region

The elderly are particularly prone to injuries in thecervicocranial region. The high incidence of atlanto-axial fractures reflects senescent changes, such as osteo-penia (especially to the base of the dens), spondyloticimmobilization of segments of the lower portion ofthe spine, and alteration of supporting soft tissues(eg, ligaments, disks, muscles) (Fig 1).

Occipital Condyles

The occipital condyle, a developmental unit of thecervicocranial region, acts as a functional part of theupper portion of the cervical spine (and is often re-ferred to as C0). Occipital condyle fractures are rare,being found at postmortem examination in 1%–5%of patients who had sustained trauma to the cervicalspine and head (23). Clinical manifestations of oc-cipital condyle fractures are highly variable, and suchfractures are not typically shown with conventionalradiography. Because as many as 30% of occipitalcondyle fractures are biomechanically unstable (dis-placed > 3 mm), their presence must be excluded inall symptomatic elderly patients who have experi-enced trauma to the head and neck. Computed to-mography (CT) is the diagnostic standard for occipi-tal condyle fractures, and the base of the skull shouldbe included in all CT examinations of the upper por-tion of the cervical spine.

In our experience, occipital condyle fractures inelderly patients are often part of multilevel fracturesof the functional craniocervical unit (Fig 2). In frac-tures of cervical vertebra C1 or combined fractures of

Figure 2. Multilevel injury in cervicocranial region, including occipital condyle fracture and "low"(Anderson-D’Alonzo type III) odontoid fracture in a 65-year-old man who fell out of top bunk on acruise ship. (a) Transverse CT scan obtained at skull base shows occipital condyle fracture (arrow),with fragment displaced medially from inferomedial aspect of right occipital condyle. (b) TransverseCT scan obtained at level of cervical vertebra C2 shows fracture line (arrows) from "low" (type III)odontoid fracture extending into body of axis.

Lom

osc

hit

z et

al

154

cervical vertebrae C1 and C2, a direct search for associ-ated fractures of the occipital condyle is indicated.

Atlas (Cervical Vertebra C1)

Among the elderly, an isolated fracture of the atlasis rare (17). Approximately 90% of the fractures ofcervical vertebra C1 occur in combination with C2fractures, typically dens fractures (Fig 3). Becausemultilevel fractures (cervical vertebrae C1 and C2) areconsidered biomechanically unstable, a heightenedbidirectional search for contiguous fractures is critical(find C1 and then look for C2, and vice versa).

Axis (Cervical Vertebra C2)

Epidemiologic comparisons between elderly andyoung populations of patients with blunt injury showstriking differences in the incidence, type, and distribu-tion of fractures. In elderly and young adult popula-tions, cervical vertebra C2 is the most frequently in-jured vertebra. In the elderly, however, the incidence isnearly twice that in younger individuals (40%–48% vs22%–28%) (12,21). Combined injuries of the motionsegment of cervical vertebrae C1 and C2, which areusually regarded as clinically unstable, are even morecommon among the elderly (69% vs 36%) (16). Inelderly patients with a history of "low-energy" injurymechanisms (fall from standing or seated height), theproportion of C2 fractures is even higher (as much as80%) (12).

Odontoid fractures are by far the most commonfractures of cervical vertebra C2. Fractures of the densaccount for two-thirds of all C2 fractures in the elderlyand are evenly distributed between “high” odontoidfractures (Anderson-D’Alonzo fracture type II) and“low” odontoid fractures (Anderson-D’Alonzo fracturetype III) (24) (Fig 4). Anderson-D’Alonzo type I odon-toid fractures are extremely rare in elderly patients. Trau-matic spondylolisthesis (hangman fractures) and verte-bral body fractures, including hyperextension teardropfractures (Fig 5), are also less commonly seen.

Lower Cervical Spine (C3 through C7)

Because of the morphologic similarities among themotion segments of cervical vertebrae C3 through C7,the fracture patterns are stereotypic. However, becauseof individual-specific degenerative conditions, includ-ing severity and in-segment degenerative transitionphysiology, there are differences in the distribution offractures. The most frequently injured motion seg-ment of the lower portion of the cervical spine is cer-vical vertebrae C5 and C6, for the elderly and youngadults alike. The motion segment of C5 through C6accounts for 20%–28% of the injuries among the eld-erly, compared with 40%-50% in younger adults(8,12,16). In comparison with the rate in the generalpopulation, cervicothoracic junction injuries are seenless frequently in the elderly (5%–10% vs 9%–18%)

(16,25). In the elderly, adjacent-level injury patternsare frequent, and multilevel fractures are generallyconsidered clinically unstable.

Fractures of the lower portion of the cervical spinecan be divided into (a) clinically stable and relativelyunimportant fractures (including transverse or iso-lated spinous process fractures, laminar fractures,avulsion fractures, and anterior compression frac-tures) and (b) clinically unstable and severe fractures(including burst fractures, hyperflexion teardrop frac-tures, and hyperextension-dislocations, which either[a] involve the facet joints with fracture and/or dis-traction or [b] are complex fractures) (Fig 6). Fracturesseen in elderly patients tend to be severe and clinicallyunstable types (55%–75% of the cases) and seem tobe independent of the causative mechanism of trauma(ie, low- vs high-energy injury mechanism) (16).

Thoracic and Lumbar Spine

Particularly in the elderly patient, the cause of acompression fracture of a thoracic or lumbar vertebracan be uncertain, and the question of metastatic re-placement of the vertebra, with subsequent collapse,versus osteopenic compression fracture is often raised.In particular, elderly patients with a compression frac-ture of a thoracic or lumbar vertebra may presentwithout a history of substantial trauma.

Besides comparison with remote radiographs (thatare often not available), certain other features maybe used as reliable predictors of a benign cause. If the

Figure 3. Atlantoaxial injury in a 74-year-old man injured as anunrestrained passenger in motor vehicle accident. Lateral radio-graph shows "low" (type II) odontoid fracture (short black arrow)and fracture through spinous process of cervical vertebra C2(white arrow), as well as bilateral fracture through the posteriorarch of atlas (long black arrows) caused by hyperextensionmechanism. Note prevertebral soft-tissue swelling (arrowheads).

Spin

e Traum

a in th

e Elderly

155

patient has sustained trauma and the fracture hassharp margins, a benign cause is likely. Another radio-graphic feature that is a reliable predictor of a benigncause for a fracture is gas within the fracture or withinthe adjacent disk space. When conventional radio-graphs show that only the anterior column is involvedin a compression fracture, no additional imagingstudies are needed. Normally, the posterior edge ofeach vertebral body is noted as a vertical line, called

the posterior vertebral body line (26). This line is nor-mally mildly convex anteriorly. Nonvisualization or aposterior convexity of this line must raise the questionof retropulsion of a fragment into the spinal canal.When there is concern that a fracture extends to theposterior longitudinal ligament or that retropulsionhas resulted in a bony fragment in the canal, furtherevaluation becomes necessary, which is usually per-formed with CT (Fig 7).

If there is a clinical suspicion of metastatic disease,further evaluation with bone scintigraphy or MR im-aging may be useful. Bone scintigraphy may help tocharacterize other lesions, and MR imaging may helpto define the morphologic structure of the fracture.Although in acute osteoporotic compression fractures,the vertebra may show signal intensity abnormalitycaused by bone marrow edema, there are certain dis-tinctive features that imply a malignant cause, includ-ing complete replacement of the vertebral body by ab-normal signal intensity, extension of abnormal signalintensity into the pedicle, epidural extension of abnor-mal signal intensity, and a paraspinal mass (27,28).Usually, bone marrow edema from an acute fractureresolves within 6 weeks, whereas changes caused bymalignant disease persist or even progress.

IMAGING TECHNIQUES TO EVALUATE SPINEINJURY IN ELDERLY PATIENTS

The optimal imaging strategy depends on the prob-ability of injury in that individual. Indications for di-agnostic imaging of elderly victims of spine traumaare widely agreed upon and include one or more ofthe following: (a) acute myelopathy or radiculopathy,(b) posterior midline tenderness, and (c) lack of a reli-able clinical examination (inability to perform clinicalexamination because of preexisting or posttraumatic

Figure 4. "Low" (type III) odontoid fracture in a 74-year-old woman injured in a fall from standing position. (a) Transverse CT scanshows "low" (type III) odontoid fracture with fracture line (arrowheads) crossing through superior portion of axis body caudad to thejunction of base of dens and axis body. (b) Coronal and (c) sagittal CT reformations more clearly show orientation of fracture line(arrowheads) relative to (b) cervical vertebra C2 lateral mass articulations (arrows) and (c) vertebral body.

Figure 5. Hyperextension teardrop fracture of axis in a 69-year-old woman injured as an unrestrained driver in motor ve-hicle accident. Lateral radiograph, obtained after placement oftongs, shows large triangular fragment (∗) comprising anteriorinferior corner of axis avulsed by intact anterior longitudinalligament during hyperextension.

Lom

osc

hit

z et

al

156

confusion or somnolence, intoxication, or distractinginjury) (21).

As in the general population, the role of imagingin acute spinal injuries in elderly patients is adjunc-tive to detection of clinical instability (9). Contro-versy exists, however, with regard to the optimal im-aging modality in victims of blunt trauma who aresuspected of having spine injury. The standardmethod for evaluating the spine in trauma is theconventional radiographic series, usually consistingof anterior, lateral, and odontoid views for the cervi-cal spine and anterior and lateral views for the tho-racic and lumbar spine. However, obtaining ad-equate views may be difficult in the acute trauma set-ting, particularly in elderly patients, and may lead toprolonged emergency department stays.

The results of a recent analysis revealed that 15%–40% of cervical spine fractures in the elderly areoverlooked in the initial survey, compared with 4%in patients younger than 65 years (11). Missed inju-ries were attributed to (a) difficulty in visualizingminimally displaced fractures in osteoporotic bone,(b) alterations in bone anatomic structures caused byspondylosis and spondylolisthesis, and (c) “satisfac-tion of search” because of the presence of multipleless important findings.

Several authors have advocated CT “screening” forcervical spine fracture in high-risk patients, and suchCT is now performed routinely at many trauma cen-ters (15,25,29–31). The results of cost-effectivenessanalysis demonstrate that CT of the cervical spine iscost-effective in patients undergoing head CT with a

Figure 6. Hyperextension-dislocation of lower portion of cervical spine (cervical vertebra C6) and incomplete spinal cord injury inan 83-year-old woman injured as an unrestrained driver in high-speed motor vehicle accident. (a) Lateral radiograph shows severedegenerative changes, with skeletal hyperostosis and bridging anterior osteophytes. At C6 level, a broad lucent line represents frac-ture crossing osteopenic spine (arrowhead). (b) Sagittal CT reformation shows anterior widening of hyperextension fracture at C6level (arrows). (c) Sagittal short inversion time inversion-recovery magnetic resonance (MR) image shows triangular area of abnor-mal high signal intensity (∗) at the C6 level, representing hyperextension injury and extensive soft-tissue swelling, with signal inten-sity alterations consistent with prevertebral hemorrhage (arrowheads).

Figure 7. Back pain with noneurologic deficit in a 74-year-old man injured in a recent fallfrom flight of stairs. (a) Trans-verse CT scan obtained at levelof lumbar vertebra L1 revealstwo-column burst fracture withretropulsion of fragment of pos-terior vertebral cortex into spinalcanal. Gas within vertebral bodyis suggestive of presence ofKümmel disease (posttraumaticosteonecrosis). (b) Sagittal CTreformation of thoracolumbarspine shows reduction of ap-proximately 30% in vertebralheight of L1.

Spin

e Traum

a in th

e Elderly

157

greater than 4% probability of cervical spine fractures(32).

Factors such as severe closed head injury, high-energymechanism, and neurologic deficits have been associ-ated with an increased risk of cervical spine fracture inthe general population (25,30). In a recent study of aclinical prediction rule for cervical spine fracture ap-plied to victims of blunt trauma aged 65 years andolder, investigators identified predictors similar tothose in the general adult population (30,33). How-ever, because of the higher proportion of fractures fromlow-energy trauma, more fractures are missed with theprediction rule in the elderly than in the general adultpopulation. To date, we are unaware of any validatedclinical prediction rules to guide the selection of pa-tients with blunt-force trauma for imaging of thoraco-lumbar spine injuries. However, it seems reasonablethat the enunciated clinical prediction rules for cervicalspine imaging may be extrapolated to the thoracic andlumbar spines, with inclusion of particular high-riskfindings (eg, “lap-belt sign”) (34). CT is advocated forpatients with a high probability of injury (30).

Further, conventional radiographic evaluation alonemay be inadequate in severely traumatized patients orfor regions with incomplete radiologic assessment withconventional radiography (eg, cervicothoracic junc-tion) (25,29). Senescent morphologic changes, such asspondylosis, skeletal hyperostosis, and paravertebralligamentous ossification, may obscure radiographicsigns of trauma on conventional radiographs.

Therefore, in the elderly trauma patient with unre-markable radiographic findings, complaints aboutpersistent pain in the posterior part of the neck, re-gardless of the injury mechanism, indicate that CT orMR imaging should be performed to exclude occultinjuries of the cervical spine. The superiority of MRimaging in soft-tissue depiction makes it the diagnos-tic standard for evaluation of spinal cord injury. Inaddition, MR imaging shows prevertebral hemorrhageand traumatic disk and facet-joint abnormalities, aswell as injury to the paravertebral ligaments, whichmay be subtle or even occult radiographically (35).

In a manner analogous to young children, manymore elderly individuals sustain acute neurologic com-promise without acute conventional or CT imaging ab-normalities, an adult variant of spinal cord injury with-out radiographic abnormality that is associated withacquired narrowing of the spinal canal and spondylosis(14,16,36). MR imaging is performed to evaluate thespinal cord and the musculoskeletal axis, particularlythe posterior ligament complex. Further, epidural pro-cesses, such as hematomas, may be readily assessed andmay guide surgical decompression. Spinal cord imag-ing is best performed with a T1-weighted MR sequenceto evaluate the size of the cord and a T2-weighted MRsequence to evaluate areas of edema and hemorrhagewithin the cord. In addition, particularly in the assess-

ment of trauma patients, T2-weighted MR sequenceswith elimination of fat signal intensity (ie, either a T2-weighted sequence with a chemical fat saturation pulseapplied or a short inversion time inversion-recovery se-quence) are a highly accurate method of evaluatingligamentous destabilizing injury (36).

In conclusion, factors that may contribute to missedinjuries in the elderly (a) include failure to identify el-derly trauma patients at risk of injury and subsequentfailure to obtain diagnostically adequate imaging stud-ies, (b) difficulty in interpreting radiographic images ofosteopenic and senescent anatomic structures, and(c) failure to appreciate the spectrum of injury amongthe elderly. The elderly are particularly prone to injuriesat the atlantoaxial complex. Compared with youngerpatients, the elderly experience injuries in the lowerportion of the cervical spine less often. When present,fractures are often severe and highly unstable and usu-ally involve more than one level. Age, in and of itself, isnot an indication for spine radiography. However, inju-ries to the spine, particularly fractures of the atlanto-axial complex, must be excluded in older patients withneck or back pain after even minor injury.

References1. Hazzard WR, Burton JR. Health problems in the elderly. In:

Braunwald E, Isslebacher KJ, Petersdorf RG, eds.Harrison’s principles and practice of internal medicine. 11thed. New York, NY: McGraw-Hill, 1987; 450–451.

2. Flemming AW, Lindner JE. Traumatic injuries. In:Yoshikawa TT, Norman DC, eds. Acute emergencies andcritical care of the geriatric patient. New York, NY: Dekker,2000; 463–488.

3. Baker SP, O’Neill BO, Ginsburg MJ, Li G. The injury factbook. New York, NY: Oxford University Press, 1992; 62–63.

4. Bialas M, Stone M. Osteoporosis. In: Pathy MSJ. Principlesand practice of geriatric medicine. Chichester, United King-dom: Wiles, 1998; 1225–1227.

5. Hobbs F, Stoops N. Demographic trends in the 20th cen-tury. Census 2000 special report CENSR-4. Washington,DC: U.S. Census Bureau, 2002; 49–70.

6. Berkowitz M. Assessing the socioeconomic impact of im-proved treatment of head and spinal cord injuries. J EmergMed 1993; 11(suppl 1):63–67.

7. Fine PR, Kuhlemeier KV, DeVivo MJ, Stover SL. Spinalcord injury: an epidemiologic perspective. Paraplegia 1979;17:237–250.

8. Hu R, Mustard CA, Burns C. Epidemiology of incident spi-nal fracture in a complete population. Spine 1996; 21:492–499.

9. White AA, Panjabi MM. Clinical biomechanics of the spine.Philadelphia, Pa: Lippincott, 1990.

10. Riggins RS, Kraus JF. The risk of neurologic damage withfractures of the vertebrae. J Trauma 1977; 17:126–133.

11. Mann F, Kubal W, Blackmore C. Improving the imaging di-agnosis of cervical spine injury in the very elderly: implica-tions of the epidemiology of injury. Emerg Radiol 2000; 7:36–41.

12. Lomoschitz FM, Blackmore CC, Mirza SK, Mann FA. Cervi-cal spine injuries in patients 65 years old and older: epide-miologic analysis regarding the effects of age and injurymechanism on distribution, type, and stability of injuries.AJR Am J Roentgenol 2002; 178:573–577.

Lom

osc

hit

z et

al

158

13. Sterling DA, O’Connor JA, Bonadies J. Geriatric falls: injuryseverity is high and disproportionate to mechanism. JTrauma 2001; 50:116–119.

14. Spivak JM, Weiss MA, Cotler JM, Call M. Cervical spine in-juries in patients 65 and older. Spine 1994; 19:2302–2306.

15. Hoffman JR, Wolfson AB, Todd K, Mower WR. Selectivecervical spine radiography in blunt trauma: methodology ofthe National Emergency X-Radiography Utilization Study(NEXUS). Ann Emerg Med 1998; 32:461–469.

16. Daffner RH, Goldberg AL, Evans TC, Hanlon DP, Levy DB.Cervical vertebral injuries in the elderly: a 10-year study.Emerg Radiol 1998; 5:38–42.

17. Lomoschitz FM, Blackmore CC, Stadler A, Linau KF,Mann FA. Fractures of the atlantoaxial complex in the el-derly: assessment of radiological spectrum of fracturesand factors influencing imaging diagnosis. Rofo FortschrGeb Rontgenstr Neuen Bildgeb Verfahr 2004; 176:222–228. [German]

18. Brandser EA, El-Khoury GY. Thoracic and lumbar spinetrauma. Radiol Clin North Am 1997; 35:533–557.

19. Bensch FV, Kiuru MJ, Koivikko MP, Koskinen SK. Spinefractures in falling accidents: analysis of multidetector CTfindings. Eur Radiol 2004; 14:618–624.

20. Ritzel H, Amling M, Posl M, Hahn M, Delling G. The thick-ness of human vertebral cortical bone and its changes inaging and osteoporosis: a histomorphometric analysis ofthe complete spinal column from 37 autopsy specimens. JBone Miner Res 1997; 12:89–95.

21. Ngo B, Hoffman JR, Mower WR. Cervical spine injury in thevery elderly. Emerg Radiol 2000; 7:287–291.

22. Weller SJ, Malek AM, Rossitch E Jr. Cervical spine frac-tures in the elderly. Surg Neurol 1997; 47:274–280.

23. Leone A, Cerase A, Colosimo C, Lauro L, Puca A, MaranoP. Occipital condylar fractures: a review. Radiology 2000;216:635–644.

24. Anderson LD, D’Alonzo RT. Fractures of the odontoid pro-cess of the axis. J Bone Joint Surg Am 1974; 56:1663–1674.

25. Hanson JA, Blackmore CC, Mann FA, Wilson AJ. Cervicalspine injury: a clinical decision rule to identify high-risk pa-tients for helical CT screening. AJR Am J Roentgenol 2000;174:713–717.

26. Daffner RH, Deeb ZL, Rothfus WE. The posterior vertebralbody line: importance in the detection of burst fractures.AJR Am J Roentgenol 1987; 148:93–96.

27. An HS, Andreshak TG, Nguyen C, et al. Can we distinguishbetween benign versus malignant compression fractures ofthe spine by magnetic resonance imaging? Spine 1995; 20:1776–1782.

28. Jung HS, Jee WH, McCauley TR, Ha KY, Choi KH. Dis-crimination of metastatic from acute osteoporotic compres-sion fractures with MR imaging. RadioGraphics 2003; 23:179–187.

29. Nunez DB, Ahmad AA, Coin CG, et al. Clearing the cervicalspine in multiple trauma victims: a time-effective protocolusing helical computed tomography. Emerg Radiol 1994; 1:273–278.

30. Blackmore CC, Emerson SS, Mann FA, Koepsell TD. Cervi-cal spine imaging in patients with trauma: determination offracture risk to optimize use. Radiology 1999; 211:759–765.

31. Stiell IG, Wells GA, Vandemheen KL, et al. The CanadianC-spine rule for radiography in alert and stable trauma pa-tients. JAMA 2001; 286:1841–1848.

32. Blackmore CC, Zelman WN, Glick ND. Resource costanalysis of cervical spine trauma radiography. Radiology2001; 220:581–587.

33. Bub LD, Blackmore CC, Mann FA, Lomoschitz FM. Cervicalspine fractures in the elderly: a clinical prediction rule. Radi-ology (in press).

34. Mann FA, Cohen WA, Linnau KF, Hallam DK, BlackmoreCC. Evidence-based approach to using CT in spinal trauma.Eur J Radiol 2003; 48:39–48.

35. Katzberg RW, Benedetti PF, Drake CM, et al. Acute cervi-cal spine injuries: prospective MR imaging assessment at alevel 1 trauma center. Radiology 1999; 213:203–212.

36. Cohen WA, Giauque AP, Hallam DK, Linnau KF, Mann FA.Evidence-based approach to use of MR imaging in acutespinal trauma. Eur J Radiol 2003; 48:49–60.

159

Imaging of ThoracolumbarSpine Trauma1

More than 10.000 spinal cord injuries occur in the United States each year, and aboutone-third are complete, resulting in paraplegia or quadriplegia. These injuries are usu-ally the result of motor vehicle accidents or falls from a height, and they often involveyoung individuals. The cost for treating spinal cord injuries in the United States is esti-mated to be around $2 billion annually. Because of the serious consequences of miss-ing an unstable spinal fracture, the injured spine should ideally be “cleared” withinthe first few minutes after the patient is admitted to the emergency department (1).

The introduction of the multi–detector row computed tomographic (CT) scannerinto the emergency department has revolutionized imaging protocols for patients withmultiple trauma, where time is of the essence. The advantages of multi–detector rowCT include speed, increased coverage, isotropic imaging, and ease of image interpreta-tion. Isotropic imaging provides uniform spatial resolution in all directions (x, y, and z),which allows the creation of two-dimensional multiplanar reformations in any arbitraryplane, as well as the creation of high-quality three-dimensional images. The results of anumber of studies have shown that multi–detector row CT is superior to radiographyin depicting fractures of the thoracolumbar spine and that multi–detector row CT cantotally replace radiography in severely injured patients (1).

Historically, injuries of the thoracic and lumbar spine have been lumped together.However, the anatomic structures and biomechanical properties of the different seg-ments in the thoracic and lumbar spine vary markedly. On that basis, the thoracic andlumbar spine can be divided into three segments: (a) thoracic vertebrae T1 throughT10, (b) thoracic vertebra T11 through lumbar vertebra L4, and (c) lumbar vertebra L5.

DISTINGUISHING FEATURES OF THE UPPERTHORACIC SPINE (T1 THROUGH T10)

The upper portion of the thoracic spine is the largest segment of the spine and is whereapproximately 10%–20% of all spinal fractures occur. A distinguishing anatomic fea-ture of the upper portion of the thoracic spine is the rib cage, which restricts motionand adds stiffness and stability to the spine (2). Some authors consider the rib cage asan integral part of the upper thoracic spine, calling it the fourth column, because of itscapacity to absorb kinetic energy during accidents. These properties are lost after ribfractures or costovertebral fracture dislocations occur in the upper thoracic spine, andits inherent stability becomes questionable (2).

Georges Y. El-Khoury, MD

1From the Department of Radiology, University of Iowa Carver College of Medicine, University of Iowa Hospitals andClinics, 200 Hawkins Dr, Iowa City, IA 52242 (e-mail: george-el-khoury@uiowa.edu).

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 159–167.

El-K

ho

ury

160

Another distinguishing feature of the upper portionof the thoracic spine relates to the considerable energyrequired to produce a fracture or fracture dislocations;therefore, the possibility of a noncontiguous spinalfracture should be kept in mind. The search for noncon-tiguous fractures requires imaging of the entire spine.

Noncontiguous fractures are commonly underdiag-nosed at conventional radiography. They are reportedto occur in 5%–20% of the patients with spinal frac-tures (Fig 1). However, after the introduction of mag-netic resonance (MR) imaging, it became obvious thatboth contiguous and noncontiguous injuries are muchmore common than previously thought. At MR imag-ing, bone bruises are counted as microtrabecular frac-tures; this creates problems for surgical planning be-cause the importance of bone bruises is not yet settled.If only conservative therapy is considered, bone bruiseshave no clinical importance; however, if surgical stabi-lization is required, then it is not known whether bonebruises are a contraindication to instrumentation (3).

Injuries of the upper portion of the thoracic spineare less common than those of the thoracolumbarjunction, and they are more likely to cause neurologicdeficits. Sixty-three percent of the patients with upperthoracic spine fractures present with a neurologic defi-cit, and the deficit is frequently complete (4). On theother hand, only 2% of the patients with lumbar spineinjuries and 32% of the patients with cervical spine in-juries have a complete neurologic deficit. Spinal corddamage in the upper thoracic spine is common becausethe ratio of the canal size to the cord size is small, andthe blood supply to the midthoracic cord is tenuous (2).

NORMAL ANATOMY AND MIMICSOF VERTEBRAL FRACTURES

Ribs articulate with vertebrae at two sites; each rib headarticulates with two adjacent vertebrae at the interver-tebral disk level, while the rib tubercle articulates withthe transverse process at the costotransverse joint (5).In the upper thoracic spine, the facet joints are ori-ented in the coronal plane, providing marked resis-tance to anterior translation.

The anteroposterior length of the vertebral bodiesgradually increases from vertebrae T1 through T12,while the transverse diameter decreases from vertebraeT1 through T3 and then progressively increases fromvertebrae T4 through T12. The height of the vertebralbodies is about 2–3 mm less anteriorly than posteri-orly. In addition, some vertebral bodies show what itis known as physiologic wedging, which is most pro-nounced in the lower portion of the thoracic spineand is especially common in male patients. In the set-ting of trauma, physiologic wedging can be confusedwith compression fractures. The wedging ratio is mea-sured by dividing the height of the anterior portion ofthe vertebral body by the height of the posterior por-

tion. Values of 0.80 in male patients and 0.87 in fe-male patients at the level of vertebrae T8 through T12are considered within normal limits (6) (Fig 2a).

Another mimicker of vertebral fractures is Scheuer-mann disease, or adolescent kyphosis. In Scheuermanndisease, an abnormality of the growth cartilage resultsin weakening of the vertebral endplates, formation ofmultiple Schmorl nodes, and narrowing of the interver-tebral disks. Vertebral body growth in Scheuermanndisease is impaired, causing anterior wedging that per-sists into adulthood. This deformity in adults may beeasily confused with compression fractures (Fig 2b).

CLASSIFICATION OF UPPERTHORACIC SPINE INJURIES

Fractures of the upper portion of the thoracic spine donot neatly fit into the Denis classification (7), whichis intended for thoracolumbar junction injuries. Mostinjuries occur in flexion and axial loading because thereis hardly any rotation in the upper thoracic spine (2).Bohlman (2) classified upper thoracic spine injuries

Figure 1. Noncontiguous thoracic spine fractures. Sagittal re-formatted CT image shows two noncontiguous vertebral frac-tures at vertebrae T5 and T10 (arrows).

Tho

racolu

mb

ar Spin

e Traum

a

161

into five types: (a) wedge compression fracture, whichis a common injury and is considered to be stable be-cause of the support provided by the rib cage; (b) sag-ittal slice fracture dislocation is also a fairly commonbut unstable injury, the basic pattern in the sagittalslice injury consisting of anterior fracture dislocationwith compression of the vertebral body below (Fig 3),and often this injury is associated with same-levelfacet joint fracture dislocations; (c) complete anteriordislocation, a rare but unstable injury; (d) posteriorfracture dislocation, also known as lumberjack para-plegia, which accounts for about 3% of all upper tho-racic spine fractures and is characterized by retrolis-thesis of the upper segment; and (e) burst fractures,which are produced by severe axial loading.

Spines that are fused as a result of ankylosing spondy-litis, diffuse idiopathic skeletal hyperostosis, or severedegenerative disk disease with bridging osteophyteshave a distinguishing injury pattern that is frequentlyassociated with a neurologic deficit (Fig 4). The injuryin this group is typically due to hyperextension andcan be caused by relatively minor trauma. Two injurymechanisms are capable of producing a hyperexten-sion fracture of the thoracic spine: (a) an anteriorimpact to the upper portion of the chest and neck, or(b) direct posterior impact to the thoracic spine. Frac-tures in this group always involve all three columns,and the spinal segments above and below the fractureact as long lever arms, rendering these injuries un-stable (8). Conventional radiographs of hyperexten-sion injury show disk space widening and retrolis-

thesis, which are hallmarks of this injury (Fig 4). Dur-ing the subacute or chronic phase, these fractures aresometimes misdiagnosed as a neoplasm or infection,especially when the history of acute trauma is lacking.

An entity that is often a source of confusion is knownas delayed posttraumatic vertebral body collapse, orKümmell disease. Patients are typically old or osteo-penic or are receiving steroid therapy. Initially, theypresent with back pain; however, the initial radiographsand CT images are negative. A few days or weeks afterthe injury, the vertebral body collapses. Some evidencesuggests avascular necrosis of the vertebral body as amechanism for the delayed collapse. Radiographs andCT images obtained after the fracture develops showintradiskal and intravertebral vacuum phenomena.

Disk herniation can occur in the upper portion ofthe thoracic spine following trauma, but compared todisk herniation in the cervical spine, it is rare. Post-traumatic disk herniations in the thoracic spine areusually associated with substantial neurologic deficit.

IMAGING

Although multi–detector row CT has been in use forabout 6 years, radiography of the spine continues tobe useful for “clearing” the upper thoracic spine inpatients who are not severely injured. Upper thoracicspine fractures are difficult to detect on chest radio-graphs, and dedicated spine radiographs are recom-mended. Adequate lateral radiographs of the thoracicspine may be difficult to obtain in patients with

Figure 2. (a) Physiologicwedging in a 21-year-old manwho was in a motor vehicle acci-dent. The anterior wedging inthoracic vertebra T12 was ini-tially thought to represent acompression fracture. This sagit-tal reformatted CT image dem-onstrates no evidence of a frac-ture. (b) Scheuermann diseasein a 42-year-old man admittedto the emergency departmentwith history of a fall. Lateral ra-diograph of the thoracic spineshows Scheuermann diseaseresembling multiple fractures(arrows).

El-K

ho

ury

162

multiple trauma because the shoulders are difficult topenetrate. A supine swimmer’s view of the upper tho-racic spine is helpful in depicting the lower cervicaland upper three thoracic vertebrae. For accurate count-ing, the C2 cervical vertebra should be included onthe lateral supine swimmer’s view. However, whencervical vertebra C2 is not included, the sternal notchcan be used as a landmark, and it is usually at thelevel of the T2 vertebral body. Lateral radiographs,

along with the anteroposterior view, are helpful in as-sessing the height of the vertebral bodies, disk spaces,endplates, and alignment of the spine. Of particularimportance is the integrity of the posterior vertebralbody line, which is normally concave anteriorly. Thespinous processes are not depicted on the lateral viewbecause the posterior ribs obscure them.

Anteroposterior radiographs are helpful for the evalu-ation of the endplates, lateral vertebral body margins,

Figure 3. Wide mediastinum and left-sided pleural effusion in a patient with sagittal slice fracture dislocation at the level of thoracicvertebrae T5 through T6. The patient presented with weakness in both lower extremities. (a) Supine anteroposterior radiograph of thechest shows a wide mediastinum (arrows) and left apical pleural capping (arrowheads). The nasogastric tube and trachea are in themidline. (b) Penetrated anteroposterior view of the thoracic spine shows the fractures in thoracic vertebrae T5 and T6 (arrow).(c) Midline sagittal T2-weighted MR image (2000/90 [repetition time msec/echo time msec]) shows the changes of a sagittal slicefracture dislocation. At the level of the injury, the cord is compressed.

Figure 4. Paraplegia occurringafter a motor vehicle accident ina patient known to have long-standing ankylosing spondylitis.(a) Lateral radiograph of the tho-racic spine shows the fracturetraversing the disk (arrow) in themidthoracic spine. The uppervertebral segment is displacedposteriorly. (b) Sagittal T2-weighted MR image (2000/90)shows that all three columns arefractured, and the cord is com-pressed between the displacedfragments.

Tho

racolu

mb

ar Spin

e Traum

a

163

and pedicles. The posterior ribs, costovertebral joints,and costotransverse joints can also be depicted. Theright paraspinal line should adhere closely to the verte-bral column, and the left paraspinal line should followthe contour of the aorta from the aortic arch to thediaphragm, albeit just lateral to the lateral margins ofthe vertebral column (5). Thickening or focal bulging ofthe paraspinal lines is a good indication of a spinal frac-ture. The spinous processes in the thoracic spine nor-mally project into the midline, and each spinous processtubercle (tip) extends slightly below the inferior end-plate of its respective vertebral body. The double spinousprocess sign seen on the anteroposterior radiograph is areliable indicator of a spinous process fracture.

Because of the advantages mentioned previously, es-pecially speed and increased coverage, multi–detectorrow CT has become an indispensable diagnostic tool inthe emergency department. For the spine, multi–detectorrow CT is particularly helpful for depicting and assessingthe severity of bone injuries. Sagittal reformations areparticularly helpful in assessing the degree of retropul-sion of bony fragments into the spinal canal (Fig 5).

MR imaging is reserved for patients with a neuro-logic deficit (9) (Fig 3). In the acute phase, MR imag-ing is useful in looking for treatable causes of the neu-rologic deficit, such as bony fragments compressing thespinal cord, disk herniation, or epidural hematoma.The ability to depict cord edema and hemorrhage helpsin predicting prognosis. Long segments of the cord in-

volved with edema or focal cord hemorrhage indicateworse prognosis (9). In the chronic phase, MR imagingcan demonstrate the sequelae of cord injury, such asmyelomalacia, syringomyelia, and cord atrophy. Liga-mentous injuries can be indirectly inferred from con-ventional radiographs or CT images; however, MR im-aging can directly show ligamentous disruption.

INDIRECT RADIOGRAPHIC SIGNS OF UPPERTHORACIC SPINE INJURIES

Indirect radiographic signs of upper thoracic spinefractures have always been important because of thedifficulty in diagnosing these fractures. These signs in-clude mediastinal widening or paravertebral hemato-ma, pleural fluid, rib fractures and costovertebral dislo-cations, the double spinous process sign, and sternalfractures. Mediastinal widening and pleural fluid,which is usually blood, are also seen with traumaticaortic rupture. Mediastinal widening is seen in morethan two-thirds of the patients with upper thoracicspine fractures above thoracic vertebra T5, and thequestion of differentiating a thoracic spine fracturefrom aortic rupture becomes a problem (10) (Fig 3).Because of the high mortality rate associated withaortic transection, it is recommended that an aorticinjury should be ruled out first in patients with a widemediastinum by using CT arteriography. In the mean-time, the spine should be immobilized and protected

Figure 5. Paraspinal hematoma caused by a burst fracture of thoracic vertebra T8. (a) Anteroposterior radiograph of the thoracicspine reveals thickening of the paraspinal stripes (arrows) and collapse of the T8 vertebral body. (b) Coronal reformatted CT imageshows the compressed T8 body and the thickened paraspinal stripes (arrows) to better advantage. (c) Sagittal reformatted CT imageshow a three-column burst fracture (arrows) with retropulsion of a posterior body fragment. The spinal canal at the level of the frac-ture is stenotic.

El-K

ho

ury

164

until it is cleared with CT (10). Spinal fractures aremuch more common than aortic ruptures, but occa-sionally the two injuries coexist.

There are two types of sternal fractures, but only onetype is associated with thoracic spinal injuries. A directsternal fracture occurs when the force applied againstthe body of the sternum displaces it posterior to themanubrium. This occurs with car accidents in whichsudden deceleration forces the anterior portion of thechest against the steering wheel. With the indirect ster-nal fracture, the excess energy from a fractured upperthoracic spine is transmitted through the ribs, displac-ing the manubrium posteriorly relative to the body ofthe sternum. Therefore a depressed, or posteriorly dis-placed, upper sternal segment indicates an indirectupper sternal fracture and should direct the attentionto a fracture in the upper thoracic spine.

INSTABILITY

Criteria for instability in upper thoracic spine traumaare not well defined. However, most surgeons wouldsurgically stabilize injuries associated with one ormore of the following findings (2): (a) a fracturedislocation, (b) posttraumatic kyphosis greater than40°, (c) indirect sternal fracture and associated ribfractures, and/or (d) costovertebral dislocations.

THORACOLUMBAR SPINE JUNCTION(T11 THROUGH L2) INJURIES

Fractures of the thoracolumbar junction are fairlycommon, accounting for about 40% of all spinalfractures. Fractures of the thoracolumbar junction arecommon because this is where the rigid upper tho-racic spine transitions to the more freely mobile

lumbar spine. The majority of fractures occur be-tween thoracic vertebra T11 and lumbar vertebra L2,while injuries below vertebra L2 are rare.

In 1983, Denis proposed the three-column con-cept (7). The anterior column consists of the anteriorlongitudinal ligament and the anterior two-thirds ofthe vertebral body and disk. The middle columnconsists of the posterior third of the vertebral body,the posterior third of the disk, and posterior longitu-dinal ligament. The posterior column consists of thepedicles, laminae, facet joints, and the posterior liga-mentous structures (facet joint capsules, ligamentumflavum, and interspinous and supraspinous liga-ments).

The reason for the enduring acceptance of thethree-column concept lies in its simplicity and abil-ity to provide a rationale for assessing fracture insta-bility. The middle column is considered pivotal inmaintaining spinal stability. The injured spine is un-able to support physiologic loads when the anteriorand middle or all three columns are compromised.The treatment of thoracolumbar fractures is still con-troversial. However, it is accepted that confirmed un-stable injuries are treated with early fusion and in-strumentation. On the basis of the three-columnconcept, Denis (7) described four basic types, eachwith subtypes, of thoracolumbar fractures. Thesetypes are compression fractures, burst fractures,flexion-distraction injuries (Chance fractures),and fracture dislocations (7).

COMPRESSION FRACTURE

A compression fracture is the most common type offracture involving the thoracolumbar junction, and itaccounts for about half of all of these fractures. Motor

Figure 6. Compression frac-ture of lumbar vertebra L1.(a) Axial CT image through theupper portion of vertebra L1shows a fracture involving theanterior column. (b) Sagittal re-formatted CT image showscompression of the superiorendplate of vertebra L1.

Tho

racolu

mb

ar Spin

e Traum

a

165

vehicle accidents and falls from a height are respon-sible for most compression fractures. Compressionfracture represents failure of the anterior columnwhile the middle column remains intact (Fig 6). Theposterior column may remain intact, or it may fail intension, depending on the magnitude of the force act-ing on the spine during the injury. The injury resultsfrom an axial load acting on a flexed spine.

On the lateral view, a compression fracture typicallyinvolves the superior endplate of the vertebral bodyproducing wedging of the vertebral body and disrup-tion of the anterior cortex just inferior to the superiorendplate. Minor compressions may be missed onaxial CT sections, and sagittal reconstructions are al-ways helpful in depicting these injuries. Initial imag-ing is performed with radiography and CT to rule outa potentially unstable burst fracture (Fig 6). Interrup-tion of the posterior body line or retropulsion of abony fragment on the lateral view should suggest aburst fracture. Widening of the interpediculate dis-tance on the anteroposterior view is also a good signof a burst fracture.

BURST FRACTURE

The burst fracture is a relatively common injury thatmakes up about 17% of all major spinal injuries.Nearly half of these fractures are associated withneurologic deficit. Burst fractures represent a failureof at least the anterior and middle columns, butmany burst fractures also involve the posterior col-umn. Most burst fractures are associated with retro-pulsion of a bony fragment, resulting in spinal canalstenosis. However, these injuries are often more seri-ous than the initial images reveal. Burst fractureshave been shown experimentally to represent a dy-namic event in which the final position of theretropulsed fragment is not representative of themaximum canal stenosis occurring during the event.In fact, maximum stenosis and maximum cord com-pression occur at the moment of the impact (11).

The lateral view shows disruption of the posteriorvertebral body line and displacement of the retro-pulsed fragment into the spinal canal (Fig 5). In themajority of cases, the retropulsed fragment is situatedat the posterior superior corner of the vertebral body,an area that is difficult to see on radiographs becauseof the overlying pedicles. The anteroposterior viewmay show the vertical fracture line through the laminaand a subtle increase in the interpediculate distance.Currently, CT images with sagittal reformations areroutinely obtained for the evaluation of burst fractures(Fig 5); patients with neurologic deficit require MRimaging. Signs of burst fracture instability include a50% or greater loss in vertebral body height, posteriorcolumn injury, progressive kyphosis, and progressiveneurologic deficit.

FLEXION-DISTRACTION INJURY(CHANCE FRACTURE)

The flexion-distraction injury accounts for about 5%of all major spinal injuries. It is produced by ahyperflexion force with the axis of rotation centeredanterior to the middle column (12). The posteriorand middle columns fail in tension, while the ante-rior column can fail in either tension or compres-sion, depending on whether the axis of rotation is ator anterior to the anterior column. If only the bonyelements of a single vertebra are involved, the injuryis believed to represent the classic Chance fracture.Often, however, the injury involves both ligamen-tous and bony structures or could even extend to anadjacent vertebra. These are referred to as the Smithfractures.

The Chance fracture horizontally splits the spinousprocess, laminae, and pedicles and extends into theposterior aspect of a single vertebral body (Fig 7);the anterior portion of the vertebral body mayshow mild compression. Radiographs frequentlyshow the fracture lines in the spinous process, lami-nae, pedicles, and posterior body. On the lateralview, there is increased height of the posterior aspectof the involved vertebral body. Because the fractureruns in the axial plane, axial CT sections may not ad-equately demonstrate these fractures. The “disap-pearing laminae” sign on axial sections should tipoff the observer to the presence of a Chance fracture.When the injury is purely ligamentous, the supra-spinous and interspinous ligaments are disrupted,along with dislocation of the facet joints and disrup-tion of the posterior longitudinal ligament and pos-terior portion of the disk. A high rate of intraabdom-inal injuries (45%) is associated with flexion-distrac-tion injuries. Neurologic deficit is seen in as many as15% of the patients with these injuries (12).

FRACTURE DISLOCATION

Fracture dislocation is a fairly common injury, ac-counting for about 20% of all major spinal injuries,and has a high incidence of associated neurologicdeficit (75%). Mechanisms implicated in producinga fracture dislocation include flexion-rotation, flex-ion-distraction, and shear forces. A fracture disloca-tion injury is characterized by displacement of onevertebral body with respect to an adjacent vertebralbody, and therefore, any horizontal translation orrotation at the level of the injury should raise thesuspicion of a fracture dislocation (Fig 8). All threecolumns fail, which results in an unstable injury.Radiographs demonstrate malalignment of the verte-bral bodies and spinous processes at the affectedlevel (Fig 8). Facet dislocation occurs with severecases. Axial CT sections demonstrate (a) the malalign-ment of the vertebral bodies, which manifests as the

El-K

ho

ury

166

“double rim” sign, or (b) the dislocated facets, whichare recognized by the presence of the “naked facets.”Sagittal, coronal, and three-dimensional reforma-tions are now routinely obtained for thorough evalu-ation of fracture dislocation anywhere in the spine.

L5 VERTEBRAL FRACTURE

Fractures of lumbar vertebra L5 are rare and oftenhave unique imaging features. The L5 vertebra is se-curely seated between the iliac wings and the sacrum,

Figure 8. Fracture dislocationat the level of thoracic vertebraeT11 through T12 in a patientwho presented with a completeneurologic deficit below vertebraT11. (a) Lateral radiograph of thethoracolumbar junction showsconsiderable anterior translationof vertebra T11 on vertebra T12.The T12 body is compressed.(b) Sagittal T2-weighted MR im-age (2000/90) shows the fracturedislocation, the cord hemorrhage(hypointense area), and cordedema (hyperintense area) at theconus. Also seen is the spinalstenosis at the site of the injuryand a small extradural hemato-ma underneath the posterior lon-gitudinal ligament. At the level ofthe injury, the ligamentum fla-vum, interspinous ligament, andsupraspinous ligament are alldisrupted.

Figure 7. Flexion-distraction(Chance) fracture. (a) Antero-posterior radiograph demon-strates the fracture traversingthe vertebral body and pedicles(arrowheads). (b) Lateral radio-graph from another patient showsa similar fracture (arrowheads).

Tho

racolu

mb

ar Spin

e Traum

a

167

and a common mechanism of injury is a high-energyaxial load. Four injury patterns have been described:(a) compression fracture, (b) burst fracture, (c) frac-ture dislocation with facet “lock,” and (d) L5 vertebralfracture associated with a complex pelvic fracture, es-pecially those that vertically split the sacrum. The lat-ter two types of injury patterns are unstable and arefrequently associated with neurologic deficit.

TRAUMATIC EPIDURAL HEMATOMA

Epidural hematomas were believed to be a rare com-plication of spinal fractures. They are reported to oc-cur in 0.5%–7.5% of spinal fractures. MR imaging isthe imaging technique of choice for detecting epiduralblood collections (Fig 8b). Epidural hematomas typi-cally originate from the epidural venous plexus and,therefore, are low-pressure collections. They werethought to represent a serious complication, causingcord compression and requiring emergent evacuation.The thinking is gradually changing, and epidural he-matomas are now believed to run a fairly benigncourse, and they rarely require evacuation, especiallyin the lumbar spine.

References1. Wintermark M, Mouhsine E, Theumann N, et al. Thora-

columbar spine fractures in patients who have sustainedsevere trauma: depiction with multi–detector row CT. Radi-ology 2003; 227:681–689.

2. Bohlman HH. Treatment of fractures and dislocations of thethoracic and lumbar spine. J Bone Joint Surg Am 1985;67:165–169.

3. Qaiyum M, Tyrrell PN, McCall IW, Cassar-Pullicino VN. MRIdetection of unsuspected vertebral injury in acute spinaltrauma: incidence and significance. Skeletal Radiol 2001;30:299–304.

4. Rogers LF, Thayer C, Weinberg PE, Kim KS. Acute injuriesof the upper thoracic spine associated with paraplegia. AJRAm J Roentgenol 1980; 134:67–73.

5. El-Khoury GY, Whitten CG. Trauma to the upper thoracicspine: anatomy, biomechanics, and unique imaging features.AJR Am J Roentgenol 1993; 160:95–102.

6. Lauridsen KN, De Carvalho A, Andersen AH. Degree of ver-tebral wedging of the dorso-lumbar spine. Acta Radiol Diagn(Stockh) 1984; 25:29–32.

7. Denis F. The three column spine and its significance in theclassification of acute thoracolumbar spinal injuries. Spine1983; 8:817–831.

8. Weinstein PR, Karpman RR, Gall EP, et al. Spinal cord in-jury, spinal fracture, and spinal stenosis in ankylosingspondylitis. J Neurosurg 1982; 57:609–616.

9. Castillo M. Current use of MR imaging in spinal trauma.Emerg Radiol 1999; 6:121–123.

10. Bolesta MJ, Bohlman HH. Mediastinal widening associatedwith fractures of the upper thoracic spine. J Bone Joint SurgAm 1991; 73:447–450.

11. Wilcox RK, Boerger TO, Allen DJ, et al. A dynamic study ofthoracolumbar burst fractures. J Bone Joint Surg Am 2003;85:2184–2189.

12. Vaccaro AR, Kim DH, Brodke DS, et al. Diagnosis andmanagement of thoracolumbar spine fractures. J Bone JointSurg Am 2003; 85:2456–2470.

NO

TES

169

Imaging of Upper ExtremityInjuries in Children1

The upper extremity is the most common site of fractures in children. Several changinganatomic features predispose to these injuries. The junction between the dense lamel-lar bone of the diaphysis and the more porous bone of the metaphysis is a site ofweakness, particularly in the distal portion of the radius. This weak transition is thesite of most of the distal radial fractures. In the distal part of the humerus, the thinplate between the olecranon fossa and the coronoid fossa constitutes the origin of thesupracondylar fractures. Finally, the transition from cartilage to bone results in in-creased susceptibility to injury. Epiphyseal separations in the distal humerus andproximal radius occur where the metaphyseal bone meets the physeal cartilage at thezone of provisional calcification. Salter-Harris–type fractures of several physes, particu-larly those of the proximal and distal humerus and proximal and distal radius, alsoarise in this region, although usually in older patients (1).

Although it is likely that the activities in which children get involved are the mainfactors predisposing to extremity injury, recent evidence shows that decreased bonedensity may be a contributing factor in some pediatric fractures. In fact, in children,there is a dose-dependent association between wrist and forearm fractures and viewingtelevision, the computer, and video games, presumably because of decreased bonedensity caused by inactivity, whereas light exercise is protective (2).

TECHNICAL CONSIDERATIONS

Radiographs of the elbow should include an anteroposterior radiograph in extensionand a lateral radiograph obtained in 90° of flexion. An appropriate study can be per-formed only if the shoulder and elbow are at the same level and if the thumb is point-ing toward the ceiling. A radiograph obtained in extension can make the posterior fatpad bulge out of the olecranon fossa and therefore can simulate an effusion (3). Whenconfronted with findings on the radiographs that are suspicious for but not diagnosticof a fracture, bilateral shallow oblique radiographs can help in making the diagnosis.Comparison with contralateral radiographs can sometimes be useful, particularlywhen the diagnostic problem relates to whether a finding is a normal variant of ossifi-cation or a fracture. However, contralateral radiographs should be used sparingly be-cause usually they are an unnecessary source of radiation.

Ultrasonography (US) of the elbow can depict unossified structures with exquisite de-tail (Fig 1). US can also be useful (a) when the question is whether there is continuity

Diego Jaramillo, MD, MPH

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 169–174.

1From the Department of Radiology, Children’s Hospital of Philadelphia, 34th St and Civic Center Blvd, Philadelphia, PA19104 (e-mail: jaramillo@email.chop.edu).

Jara

mill

o

170

between the unossified epiphysis and metaphysis (suchas would occur with a Salter-Harris fracture of the distalhumerus in an infant), (b) when there is a question ofdislocation of unossified structures, and (c) when it isimportant to detect whether the unossified epiphysishas been injured (4,5).

Magnetic resonance (MR) imaging is useful for theevaluation of subtle injuries to unossified structures.This is true of the evaluation of sporadic cases of lat-eral condylar fractures, in which MR imaging can es-tablish whether there is extension into the unossifiedepiphysis, or in avulsions of the medial epicondyle.MR imaging is also useful for the evaluation of inju-ries involving primarily ligamentous or muscularstructures. In older children, as in adults, the evalua-tion of sequelae of anterior dislocation of the shoul-der and the evaluation of wrist pathologic abnormali-ties of the triangular fibrocartilage complex and thecarpal ligaments are done primarily with MR imaging.

Computed tomography (CT) can help in the evalua-tion of subtle injuries. CT can also be used to depictthe relationship between fracture fragments in cases ofseverely comminuted or complex displaced fractures.

DEVELOPMENTAL PITFALLS THAT CANRESEMBLE INJURIES

Most developmental pitfalls in the upper extremityare found in the elbow. The sequence of appearanceof the multiple ossification centers of the elbow is asfollows: capitellum, 0–2 years; radial head, 4–5 years;medial, or internal, epicondyle, 6–7 years; trochlea,8–10 years; olecranon, 10 years; and lateral, or exter-nal, epicondyle, 11 years. This sequence can be sum-marized by the initials CRITOE, with “IT” being the

most important; if the trochlea is present, the medialepicondyle should be visible. If the trochlea is notpresent, an avulsion of the medial epicondyle intothe joint should be presumed. The fusion of the ossi-fication centers follows a different sequence, with thecapitellum, trochlea, and external epicondyle fusingtogether prior to becoming fused with the humeralmetaphysis. The trochlea mineralizes in a fragmentedfashion, with multiple ossification centers ultimatelycoalescing to form the trochlear epiphysis.

The lateral and medial epicondyles are oftenslightly separated from the distal humeral metaphy-sis and can resemble an apophyseal separation (6).In the wrist, the scapholunate interval measured ra-diographically is wider in children than in adults.This spurious separation is due to the relatively lateossification of the lower pole of the scaphoid; on MRimages, the scaphoid and lunate are never separated,regardless of age (7).

On MR images, the preossification center (the hy-pertrophic changes developing in the unossified epi-physis prior to ossification) can result in a focal areaof increased signal intensity on T2-weighted images(8). This is an issue particularly in the trochlea,where the ossification center is always fragmented,and initially only small fragments may be seen ra-diographically. It is also important to note that the

Figure 2. Epiphyseal separation in a 6-month-old male infantwho had suffered from child abuse. Radiograph shows thatphysis of proximal humerus (arrow) is wide, and humeral epi-physis is displaced medially with respect to shaft. There is peri-osteal calcification in this infant, who had an inflicted subperi-osteal hemorrhage.

Figure 1. Coronal US image of distal humeral epiphysis in a1-year-old child suspected of having a dislocation of the elbow.Articular relationships between olecranon (o) and trochlea (t)are normal. There is early ossification of capitellum (c).

Up

per Extrem

ity Inju

ries in C

hild

ren

171

insertions of the flexor and pronator tendons oftenshow high signal intensity on gradient-recalled echoMR images (9).

SHOULDER INJURIES

Physeal fractures and chronic physeal injuries of theproximal humerus are the most common injuriesduring childhood, whereas dislocations with glenoidlabral injuries become more frequent in adolescence. Inolder children, physeal fractures do not constitute a di-agnostic challenge and can be evaluated with radio-graphs. In infants and younger children, however, dis-continuity between the proximal humeral metaphysisand the proximal ossification center can be subtle. Withproximal epiphyseal injuries, the ossification center isusually displaced medially with respect to the shaft,and the physis appears unusually wide (Fig 2). This canbe seen in the context of child abuse. It is important torecognize that in the absence of acute trauma, physealwidening can result from repeated physeal stress. Thisoccurs most frequently with baseball players. Fracturesof the humeral shaft can often be greatly displaced, butmost such fractures remodel extremely well, regardlessof the initial fragment separation.

Disruptions of the glenoid labrum result from ante-rior dislocations. They are best evaluated with MR ar-thrography, but axial nonenhanced arthrography is a

useful diagnostic modality. It is important to recog-nize that in younger children and adolescents, the gle-noid may not be fully ossified, such that there is aband of high-signal-intensity cartilage between thelow signal intensity of the glenoid fibrocartilage andthe bony glenoid.

ELBOW INJURIES

Most fractures of the elbow result from hyperexten-sion-rotation with valgus or varus stress. The mecha-nism of injury separates elbow fractures into threemain groups and explains the findings associatedwith some of these injuries: (a) Hyperextension withvertical stress results in supracondylar fractures, lon-gitudinal linear ulnar fractures, and buckle fractures;(b) hyperextension with valgus stress results in im-paction fractions of the radial head and neck, trans-verse olecranon fractures, and medial epicondylarfractures; and (c) varus stress produces Monteggiafractures, lateral condylar avulsion fractures, trans-verse olecranon fractures, and longitudinal linear ul-nar fractures (10).

Supracondylar fractures account for nearly 60% ofthe elbow injuries in children (11). When subtle, su-pracondylar injuries constitute a diagnostic challenge.The fracture occurs through an area of weakness in themetaphysis of the distal humerus, where the olecra-non and coronoid fossae are separated by a thin, oftenporous bony plate. Almost invariably, the distal frag-ment is displaced posteriorly. Radiographic diagnosisof subtle supracondylar fractures is based on (a) thedetection of an elbow effusion and (b) signs of poste-rior displacement of the distal fragment. The diagno-sis of an elbow fracture is established by elevation ofthe anterior fat pad or by posterior displacement ofthe normally invisible posterior fat pad, which ren-ders it detectable on the lateral radiograph. When thedistal humerus is posteriorly displaced, a line throughthe anterior cortex of the humerus (the anterior hu-meral line) intersects the anterior third of the capitel-lum instead of bisecting the capitellum.

The second most important injury, the lateralcondylar fracture, accounts for 15% of pediatric el-bow injuries. This fracture results from a varus stresson the elbow, and the fracture line extends from themost distal lateral humeral metaphysis, across thegrowth plate, and into the epiphyseal cartilage. Insome cases, the fracture involves the articular surface,in which case the fracture is unstable. Otherwise, thefracture can stop in the epiphyseal cartilage. The di-agnosis is difficult when only a small sliver of meta-physeal bone is separated from the parent bone; inthis case, oblique radiographs can show the fragmentto better advantage. MR imaging can be useful to de-termine whether the fracture involves the articularsurface (Fig 3).

Figure 3. Lateral condylar fracture in a 4-year-old boy. Coro-nal fat-suppressed fast spin-echo T2-weighted MR image of dis-tal humerus shows fracture extending along low-signal-intensitydistal humeral epiphysis. Fracture stops before articular surface(arrow).

Jara

mill

o

172

Other fractures of the elbow include epiphysealseparations of the distal humerus or the proximal ra-dius. These lesions usually occur in infants or youngchildren and are sometimes seen in the context ofchild abuse.

Another important elbow injury is the avulsion ofthe medial epicondyle that occurs because of theforceful contraction of the pronators and flexors in-serting on it (Fig 4). A separation greater than 5 mmindicates the need for surgery. At times, the distal frag-ment can be pulled into the joint, and the diagnosiscan be difficult.

Fractures that are not readily apparent generally arefound to be supracondylar fractures or lateral condy-lar fractures. Fractures of the radial head, a commonsource of adult occult injuries, are extremely rare inchildren. A proximal elbow injury in a child is usuallya Salter-Harris type 2 fracture of the radial neck.

Elbow dislocations occur primarily as part of thecomplex of Monteggia injuries, in which an anteriordislocation of the proximal radius is associated with amiddiaphyseal ulnar fracture (Fig 5). The isolated an-terior dislocation of the radial head that occurs withhyperextension and rotation (nursemaid elbow) isusually reduced during the flexion and external rota-tion required for radiographic positioning, and hencethis entity usually has a normal radiographic appear-ance. A line extending along the radial shaft proximal

to the radial tubercle should intersect the capitellumin every projection regardless of obliquity. Disruptionof this radiocapitellar line is useful in diagnosing un-suspected anterior radial head dislocations.

Imaging evaluation of elbow injuries relies prima-rily on anteroposterior and lateral radiographs. Addi-tional imaging can include oblique radiographs. Con-troversy exists regarding the predictive value of elboweffusions in the context of trauma. Donnelly et al (12)have suggested that an elbow effusion is only seen in17% of all elbow injuries, but if the effusion is persis-tent, a fracture is much more likely. On the basis ofcomparison with MR images, Major and Crawford(13) suggested that elbow effusions in children are as-sociated with fractures in more than half of the cases.Our own experience with multi–detector row CT sug-gests that occult fractures are indeed prevalent in chil-dren with elbow effusions (unpublished data) (Fig 6).Comparison with the contralateral side is rarely help-ful but may assist in the differentiation between a frac-ture and a normal developmental variant. CT helps incases of subtle injuries or in severely displaced injuriesby showing the separation of the fragments. US can beused for the evaluation of epiphyseal separations. MRimaging can assist in the determination of the extentof the cartilaginous injuries, as in some nondisplacedlateral condylar injuries (14,15).

Figure 4. Valgus injury of elbow in a 9-year-old boy. Frontalradiograph shows avulsion of medial epicondyle. Figure 5. Monteggia fracture resulting in subtle dislocation in

a 9-year-old boy. Radiocapitellar line is discontinuous becauseradial head is anteriorly dislocated with respect to capitellum.

Up

per Extrem

ity Inju

ries in C

hild

ren

173

FOREARM AND WRIST INJURIES

The radius and ulna are the most commonly frac-tured bones in childhood, accounting for 36% of allfractures before skeletal maturity (16). Fractures inthe forearm usually involve both the radius andulna. If only one bone is obviously fractured, it isimportant to search for subtle injury in the contralat-eral bone. For example, an ulnar fracture can be as-sociated with a bowing (plastic) fracture of the ra-dius. An ulnar fracture can also be associated with asubtle anterior dislocation of the proximal radius(Monteggia fracture). Conversely, an isolated frac-ture of the radius may be associated with a bowingfracture of the ulna or with a subtle distal ulnar dis-location (Galeazzi fracture). Monteggia injuries are

much more common in children than are Galeazzifractures.

Most injuries in the region of the wrist are bucklefractures of the distal radius. The fractures usuallybuckle one cortex, resulting in dorsal or volar angula-tion with minimal displacement. These fractures canbe subtle, particularly on the frontal view. On the lat-eral radiograph, the fractures can be easily detected asan abrupt curvature in the metaphyseal cortex. Dorsalinjuries occur more commonly than ventral ones.When the buckling of the cortex occurs circumfer-entially, the fracture is termed a torus fracture becauseit resembles the base (torus) of a column. Metaphys-eal buckle fractures have little clinical importance andrequire no imaging beyond radiography. Displace-ment of the pronator quadratus fat pad can be a use-ful indirect sign of a distal forearm fracture.

The distal radial physis is the site of 60% of allSalter-Harris–type fractures. Most injuries are type 1and 2 fractures, and they seldom result in growth dis-turbance. The infrequent posttraumatic bony bridgesusually occur in the central portion of the distal radialphysis. This is in contrast to the bony bridges found indyschondrosteosis (the genetic mesomelic dysplasiathat accounts for most cases of Madelung deformity),in which the bony bridge is on the ulnar side of thephysis. Repeated injury to the distal radial growthplate results in progressive widening and irregularityof the distal femoral physis that can resemble an acute

Figure 6. Images in a 5-year-old boy who was unable to movehis elbow after falling on outstretched hand. (a) Lateral radio-graph shows fat pads anteriorly (arrowhead) and posteriorly (ar-rows). (b) Sagittal reconstruction of elbow shows that anterior (ar-rowhead) and posterior (arrows) fat pads are displaced by elboweffusion. (c) Sagittal reconstruction at another level shows thatthere is subtle fracture (arrow) of anterior distal humeral metaphy-sis adjacent to capitellum, an occult Salter-Harris type 2 injury. Anapparent fragment anterior to ulna corresponds to partial section ofradial head.

Jara

mill

o

174

fracture. This is most commonly seen in gymnasts andresults in shortening of the radius. Injuries to the trian-gular fibrocartilage complex and the carpal bones andligaments are common in adolescents, but the injuriesare usually no different than those seen in adults.

Fractures of the ulnar styloid process occur com-monly in association with radius injuries but almostnever as isolated injuries. For this reason, when a sty-loid process fracture is identified, a distal radial frac-ture should be sought (17).

HAND INJURIES

In infants, fractures of the hands are often seen in thecontext of child abuse. These injuries are best detectedin the oblique projection and are primarily dorsalbuckle fractures of the bases of the proximal phalan-ges (18). Older children can have similar injuries asthe result of hyperextension of the fingers, often oc-curring during basketball or volleyball.

The phalanges are also common sites of physeal in-juries. These fractures are truly metaphyseal, ratherthan physeal, because typically a small sliver of bonecan be seen adjacent to the physis. Volar plate injuriesin children are less common than in adults, with asmall sliver of bone being detached from the base ofthe middle phalanx. At the base of the thumb, aSalter-Harris type 2 injury of the physis in the proxi-mal metacarpal is the counterpart of the Bennett frac-ture seen in adults. Injuries to the first metacarpopha-langeal joint (the gamekeeper or skier thumb) areidentical to those of the adult. In general, radiographsare sufficient to diagnose and define hand injuries. Inthe young, the evaluation of unossified epiphysealseparations of the phalanges can be subtle and mayrequire MR imaging.

In conclusion, imaging of injuries of the upper ex-tremity in children is challenging because of multipledevelopmental variations that can simulate diseaseand because the patterns of injury are unique. Frac-tures involving the cartilaginous structures may besubtle on radiographs. Comparison radiography andUS are used more frequently than in adults. MR imag-ing and CT can be extremely useful in certain injuries,but the roles of these modalities have not yet beenfully determined.

References1. Peterson HA. Physeal fractures of the elbow. In: Morrey

BF, ed. The elbow and its disorders. 2nd ed. Philadelphia,Pa: Saunders, 1993; 248–265.

2. Ma D, Jones G. Television, computer, and video viewing;physical activity; and upper limb fracture risk in children: apopulation-based case control study. J Bone Miner Res2003; 18:1970–1977.

3. Murphy WA, Siegel MJ. Elbow fat pads with new signs andextended differential diagnosis. Radiology 1977; 124:659–665.

4. Davidson RS, Markowitz RI, Dormans J, Drummond DS.Ultrasonographic evaluation of the elbow in infants andyoung children after suspected trauma. J Bone Joint SurgAm 1994; 76:1804–1813.

5. Graif M, Stahl-Kent V, Ben-Ami T, Strauss S, Amit Y, ItzchakY. Sonographic detection of occult bone fractures. PediatrRadiol 1988; 18:383–385.

6. Silberstein MJ, Brodeur AE, Graviss ER, Luisiri A. Somevagaries of the medial epicondyle. J Bone Joint Surg Am1981; 63:524–528.

7. Kaawach W, Ecklund K, Di Canzio J, Zurakowski D, WatersPM. Normal ranges of scapholunate distance in children 6to 14 years old. J Pediatr Orthop 2001; 21:464–467.

8. Jaramillo D, Waters PM. MR imaging of the normal devel-opmental anatomy of the elbow. Magn Reson Imaging ClinN Am 1997; 5:501–513.

9. Sugimoto H, Ohsawa T. Ulnar collateral ligament in thegrowing elbow: MR imaging of normal development andthrowing injuries. Radiology 1994; 192:417–422.

10. John SD, Wherry K, Swischuk LE, Phillips WA. Improvingdetection of pediatric elbow fractures by understanding theirmechanics. RadioGraphics 1996; 16:1443–1460; quiz1463–1464.

11. Houshian S, Mehdi B, Larsen MS. The epidemiology of el-bow fracture in children: analysis of 355 fractures, with spe-cial reference to supracondylar humerus fractures. J OrthopSci 2001; 6:312–315.

12. Donnelly LF, Klostermeier TT, Klosterman LA. Traumatic el-bow effusions in pediatric patients: are occult fractures therule? AJR Am J Roentgenol 1998; 171:243–245.

13. Major NM, Crawford ST. Elbow effusions in trauma in adultsand children: is there an occult fracture? AJR Am J Roent-genol 2002; 178:413–418.

14. Beltran J, Rosenberg ZS, Kawelblum M, Montes L, BergmanAG, Strongwater A. Pediatric elbow fractures: MRI evalua-tion. Skeletal Radiol 1994; 23:277–281.

15. Beltran J, Rosenberg ZS. MR imaging of pediatric elbowfractures. Magn Reson Imaging Clin N Am 1997; 5:567–578.

16. Lyons RA, Delahunty AM, Kraus D, et al. Children’s frac-tures: a population based study. Inj Prev 1999; 5:129–132.

17. Stansberry SD, Swischuk LE, Swischuk JL, Midgett TA.Significance of ulnar styloid fractures in childhood. PediatrEmerg Care 1990; 6:99–103.

18. Nimkin K, Spevak MR, Kleinman PK. Fractures of thehands and feet in child abuse: imaging and pathologic fea-tures. Radiology 1997; 203:233–236.

175

High-Energy Blunt-ForceInjuries to the

Upper Extremity1

UPPER EXTREMITYIMAGING IN THESEVERELY INJUREDPATIENT

Radiologic examination of theupper extremity in the severelyinjured patient begins, oddlyenough, with the initial traumachest radiograph. The radio-graph should be carefullyexamined for injury to theclavicles and shoulders, in ad-dition to the traditional evalu-ation of the heart, mediasti-num, lungs, and ribs. Fracturesof the clavicle, sternoclaviculardislocation, acromioclavicularseparation, fractures of thescapula and shoulder joint,glenohumeral dislocation, and scapulothoracic dissociation are all injuries that can bediagnosed or suspected from the findings on the trauma chest radiograph.

At Harborview Medical Center, we performed an internal quality assurance review offindings on the trauma chest radiograph that were missed by residents and fellows. Wefound that fractures of the clavicle and glenohumeral dislocation were by far the mostfrequently missed findings (Fig 1), which confirms the importance of examining theupper extremity on this study.

Conventional radiographs of the upper extremity are obtained after the initialclinical examination. They are frequently suboptimal because of the injuries to thepatient, and additional views are often obtained to further evaluate the region inquestion. If the condition of the patient is clinically stable, computed tomography(CT) of the upper extremity is used to assist the orthopedic surgeon in planning forfracture management and to clarify confusing or complex injuries. If a dislocationis present, CT is usually performed after closed reduction. Sagittal and coronal

Thurman Gillespy III, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 175–186.

1From the Department of Radiology, University of Washington, Harborview Medical Center, Box 359728, 325 Ninth Ave,Seattle, WA 98104-2499.

Figure 1. Upper extremity injury on trauma chest radiograph.The left shoulder is dislocated anteriorly (arrow).

Gill

esp

y

176

reformations are often helpful in demonstratingfracture fragment relationships. Three-dimensionalreformations can be used for further clarification,but they are not a routine part of our practice.

Magnetic resonance (MR) imaging is excellent indelineating soft-tissue structures, especially tendonsand ligaments of the joint, but emergency MR imag-ing is usually reserved for the brain and spine. Weseldom use MR for upper extremity imaging in the

severely injured patient because most displacedfractures require open reduction and internal fixa-tion, where the joint in question can be directly in-spected.

Conventional angiography of the upper extremityis performed whenever the physical examination ora particular fracture pattern indicates that a vascularinjury is likely. We do not use MR angiography forthe evaluation of extremity trauma.

Figure 2. Sternoclavicular dislocation. (a) Trauma chest radio-graph shows widening of right superior mediastinum (arrow-heads), consistent with mediastinal hematoma. (b) Axial CT im-age of chest shows widening of right sternoclavicular joint (ar-row) and inferior displacement of right proximal clavicle. Smallsliver of bone (arrowhead) at proximal right clavicle is proximalclavicle epiphysis. (c) CT angiogram shows hematoma in supe-rior mediastinum (arrows). Filling defect in right brachiocephalicartery (arrowhead) indicates vascular injury. (d) Axial CT imageof chest shows abnormal air collection (arrow) adjacent to tra-chea, which suggests tracheal laceration. (e) Angiogram showsdissection of right brachiocephalic artery (arrows).

Hig

h-En

ergy In

juries to

the U

pp

er Extremity

177

BRACHIAL PLEXUS INJURY

Injury to the brachial plexus can cause devastating neu-rologic compromise to the upper extremity. The mecha-nism of injury includes direct trauma to the head andneck region (both blunt-force and penetrating injuries)and violent abduction of the neck and shoulder (oftenfrom motorcycle accidents). Clinical evaluation of theinjury includes distinguishing between preganglionicinjuries (which never heal) and postganglionic injury(which may heal). MR imaging of the neck and axilla isoften used for staging brachial plexus injury.

STERNOCLAVICULAR DISLOCATION

Most injuries to the clavicle are not associated with seri-ous complications, but sternoclavicular dislocation cancause life-threatening injuries in the mediastinum (Fig2). Sternoclavicular dislocation is uncommon, account-ing for perhaps 2%–3% of all shoulder girdle disloca-tions. The injury is important because of commoncomplications, and delays in diagnosis are common.

The clavicle is the only fixed attachment of the upperextremity to the trunk. With indirect anterior or poste-rior blows to the shoulder, the clavicle acts as a fulcrumthrough the coracoclavicular ligaments. Anterior force tothe shoulder causes an anterior sternoclavicular disloca-tion, and posterior force to the shoulder causes poste-rior sternoclavicular dislocation. The classic mechanismis the “pile-on” injury in American football and rugby.

Sternoclavicular dislocation most often occurs inlate adolescence and young adulthood. Of interest,the epiphysis of the proximal clavicle is the last toform (age, 18–21 years) and the last to fuse (age, 25years). Many sternoclavicular dislocations in youngadults are probably epiphyseal injuries.

Anterior sternoclavicular dislocation is more com-mon but is not associated with mediastinal injury. Pos-terior dislocation, on the other hand, can cause injury

to the aorta, great vessels, trachea, esophagus, recurrentlaryngeal nerve, jugular vein, and the superior venacava. Sternoclavicular dislocation should be consideredwhen there is evidence of hematoma in the upper me-diastinum (Fig 2).

Both injuries are famously difficult to diagnose, anddelays in treatment are common. Sternoclavicular dis-location should be suspected whenever the medialclavicles are not at the same level on a trauma chestradiograph because the dislocations are often supe-rior, in addition to anterior or posterior. Anteroposte-rior (AP) views of the sternum with 40° cephalad an-gulation (Rockwood view) can be used, but conven-tional radiographs are unreliable and difficult tointerpret. CT is now the preferred imaging modality inpatients suspected of having sternoclavicular disloca-tion. Not uncommonly, the dislocation is diagnosedwhen CT angiography is obtained for evaluation ofmediastinal hematoma.

SCAPULA FRACTURE

The scapula is rarely injured except in patients withsubstantial trauma. Motor vehicle accidents probablycause about three-fourths of the cases, and 80% ofscapula fractures are associated with other injuries, es-pecially to the head and thorax. Scapula fractures canoccur in association with acromioclavicular and gleno-humeral dislocation, especially fractures of the spine,acromion, and glenoid.

Fractures are often demonstrated on the chest radio-graph or the AP view of the shoulder and can be furtherevaluated on the lateral (scapular Y) view (Fig 3). CT isused to determine if glenoid fractures are intraarticular(which may require internal fixation) and to assessmore complex injures. Three-dimensional CT recon-structions also can be helpful for understanding morecomplex injuries (Fig 4).

Figure 3. Extraarticular scapula fracture. (a) AP view of shoulder shows that fracture through body of scapula (arrowhead) extendsto neck of glenoid (arrow). (b) Scapula Y view shows scapula fracture (arrows) clearly. Glenoid is displaced anterior to body ofscapula. (c) CT image of shoulder shows that fracture (arrowheads) does not involve glenohumeral joint.

Gill

esp

y

178

There are some noteworthy anatomic variants thatshould not be confused with a scapula fracture. Theseinclude os acromiale (nonunited acromion apophy-sis, often mistaken for acromion fractures), secondaryossification center of the inferior angle of the scapula,and nutrient foramina near the neck of the glenoid.

SCAPULOTHORACIC DISSOCIATION

Scapulothoracic dissociation is a serious injury charac-terized by lateral displacement of the entire forequar-ter. This injury should be suspected clinically or radio-graphically whenever there is massive soft-tissue swell-ing in the shoulder region with no soft-tissue defect(Fig 5). The diagnosis can be made on the traumachest radiograph by noting lateral displacement of themedial scapular border, compared with the other side,without a soft-tissue defect. Lateral scapular displace-ment with a soft-tissue defect indicates a partial am-putation, a distinct and equally serious injury. Com-monly associated injuries include (a) fractures of thescapula and clavicle and (b) acromioclavicular separa-tion or sternoclavicular dislocation.

Scapulothoracic dissociation indicates injury and dis-ruption of the forequarter musculature, often associ-ated with injury to the subclavian and brachial vessels

and to the brachial plexus. Both CT and angiographyare used to assess skeletal, soft-tissue, and vasculardamage. Scapulothoracic dissociation should be distin-guished from the rare scapulothoracic dislocation,which is traumatic separation of the inferior scapulo-thoracic articulation without neurovascular injury.

HUMERUS FRACTURE

Humerus fractures are uncommon in the 20–45-year-old age group and generally indicate severe trauma(Fig 6). Proximal humerus fractures in this age groupare often associated with fracture-dislocations of theglenohumeral joint (Fig 7). The axillary neurovascularstructures are at some risk from bone fragments fromthe proximal humeral shaft or humeral head. CT is of-ten used to document fracture fragment position andto assess the articular surfaces of the humeral headand glenoid.

In any age group, fractures at the junction of theproximal two-thirds and distal one-third of the hu-meral shaft can injure the radial nerve (Fig 8). As the ra-dial nerve exits the axilla, it courses posterior and thenlateral to the humeral shaft. Two-thirds of the waydown the shaft of the humerus, the nerve is tetheredlateral to the humerus in the lateral intermuscular septa

Figure 4. Intraarticular scapula fracture.(a) AP view of scapula shows fracture throughupper scapula (arrowheads) extending towardglenoid. (b) Scapula Y view shows that fracture(arrows) extends through base of coracoid andinto upper glenoid, but extent of glenoid in-volvement is unclear. (c) CT image of shouldershows that coracoid and portion of superior gle-noid are detached as separate fragment (ar-row). (d, e) Three-dimensional CT reconstruc-tion more clearly delineates size and position ofcoracoid-glenoid fracture fragment (arrow).

Hig

h-En

ergy In

juries to

the U

pp

er Extremity

179

Figure 5. Scapulothoracic dis-sociation. (a) Trauma chest ra-diograph shows multiple left riband scapula fractures (arrow-heads) associated with massivesoft-tissue swelling (S) over leftshoulder. Medial border of rightscapula (arrows) is in normal po-sition, whereas left scapula isdisplaced lateral to chest wall.(b) CT image of chest showsmultiple fractures (arrowheads)of left scapula, associated withmassive soft-tissue swelling (S).(c) Presubtraction angiogramshows relationship of left subcla-vian and brachial arteries tofractures of scapula and ribs. (d) Subtraction angiogram showsno evidence of vascular injury.

Figure 6. Motor vehicle accident victim. (a) AP view of hu-merus shows displaced and angulated fracture in midportion ofhumerus. (b) AP view of forearm shows severity of trauma, withcomminuted fractures of midportions of radius and ulna.

and is at risk for laceration by a fracture fragment. Theradial nerve can also become entrapped in the humerusfracture at this location during closed manipulation.

ELBOW INJURIES

Injuries to the elbow, including the distal portion ofthe humerus, are generally caused by force transmittedthrough the bones of the forearm. Elbow injuries in theyoung and elderly are commonly caused by a fall onthe outstretched hand, but in young to middle-agedadults, more serious accidents involving automobiles,motorcycles, and falls from a height are common.

In the adult, fractures of the distal portion of thehumerus are typically intraarticular T- or Y-shapedtranscondylar types or their more comminuted vari-ants (Figs 9, 10). Distal humerus fractures are usu-ally readily seen on the initial radiographs, but theimages must be carefully examined for less obviousassociated fractures to the olecranon, capitellum, orradial head and for any dislocation.

Although uncommon, displaced or neglected frac-tures of the distal humerus can injure the brachial ar-tery, which lies anterior to the joint. Injury to the bra-chial artery can lead to disabling ischemic injury to

Gill

esp

y

180

Figure 8. Humerus frac-ture. AP view of humerusshows fracture at junctionof upper two-thirds andlower one-third of humer-us. No radial nerve injurywas present at physicalexamination.

Figure 7. Fracture-dislocation of right shoulder. (a) AP view of shoulder shows com-plex fracture-dislocation of glenohumeral joint. (b) Axial CT image show that largefragment is missing from humeral head (H). Glenoid is normal. (c) Oblique coronal and(d) oblique sagittal CT reformations show large portion of humeral epiphysis (arrow-heads) displaced inferiorly and trapped beneath inferior rim of glenoid (G).

the muscles of the forearm and hand, known asVolkmann contracture.

In adults, fractures of the epicondyles are usually in-traarticular. If the fracture of either the medial or lat-eral epicondyle includes the lateral aspect of the tro-chlea (Fig 11), the injury is considered mechanicallyunstable, and internal fixation is often warranted. Anyinjury to the medial epicondyle can injure the ulnarnerve, which passes through a groove on the posteriormedial epicondyle, the cubital tunnel.

Olecranon fractures are common in the elderly butalso occur in more serious trauma at any age. These in-juries are more likely to be open fractures, with grosscontamination of the wound and elbow joint (Fig 12).Combined fractures of the humeral shaft and forearmare not uncommon, which result in the grossly unstable“floating elbow” (Fig 13).

Elbow dislocations are common in both children andadults. In high-energy blunt-force trauma, elbow dislo-cations often occur in combination with dramatic frac-tures of the distal humerus or olecranon (Fig 14). In anyelbow dislocation, the radiograph must be analyzedcarefully for any associated fracture because dislocationwithout fracture is uncommon (Figs 15, 16). Soft-tissueinjuries associated with elbow dislocation include in-jury to ligaments and tendon, injury to the brachial ar-tery, and entrapment of the median nerve in the joint.

It is useful to consider the spectrum of fractures andfracture-dislocations of the proximal ulna. Three pat-terns are common: (a) fracture of the mid to proximalportion of the olecranon without dislocation (Fig 12)(classic olecranon fracture), (b) fracture of the distal por-tion of the olecranon with radial head dislocation (Fig16) (anterior fracture-dislocation of the elbow), and(c) extraarticular fracture of the proximal portion ofthe ulna with radial head dislocation (Fig 17)(Monteggia fracture-dislocation).

CT is helpful for staging the more complex elbowfractures and fracture-dislocations. Depending on theseverity of the initial injury, the CT examination is of-ten performed after closed reduction.

Hig

h-En

ergy In

juries to

the U

pp

er Extremity

181

Figure 11. Epicondyle fracture. Intra-articular fracture (arrowheads) of lateralepicondyle involves lateral portion of tro-chlea (T).

Figure 9. Comminuted T-type trans-condylar fracture of distal humerus. (a) APview shows intraarticular extension (arrow-head) of fracture. (b) Lateral view.

Figure 10. Combined T- and Y-type transcondylar fracture of distal humerus. (a) APview. (b) Lateral view.

Figure 12. Olecranon fracture.(a) Lateral view shows dis-placed fracture of olecranonwith large soft-tissue wound.Gas and foreign material extendinto elbow joint. (b) AP viewshows gas (arrowheads) withinelbow joint, which indicates jointcontamination.

Gill

esp

y

182

HAND AND WRIST

Lunate and Perilunate Dislocation

Approximately 10% of all carpal injuries are lu-nate or perilunate dislocations. Associated fracturesand ligament injury are common and generally oc-cur within the “zone of vulnerability” described byGilula. Carpal dislocations are usually obvious onthe lateral view, but important clues to carpalmalalignment can be detected with careful analysisof the posteroanterior (PA) view. On the lateralview, no carpal bone should cross either the volar ordorsal radial line (Fig 18).

The hallmark of most carpal dislocations is the dis-location of the capitate and the lunate on the lateralview. Perilunate dislocation is present when the capi-tate crosses the dorsal radial line and when the lunatelies behind the volar radial line (Fig 19). Volar tiltingof the lunate is common and should not be confusedwith a lunate dislocation. Perilunate dislocations aretwo to three times more frequent than lunate disloca-tion. Roughly 75% of the cases are accompanied by ascaphoid waist fracture, which is then termed a trans-scaphoid perilunate dislocation. The dislocation is al-most always volar.

Lunate dislocation is present when the lunatecrosses the volar radial line and when the capitatedoes not cross the dorsal radial line (Fig 20). The lu-nate is typically volar tilted, which results in theclassic pie-shaped lunate on the PA view. Most lu-nate dislocations do not have an associated fracture.

Not uncommonly, a mixed pattern of carpal dislo-cation is present, in which both the lunate and thecapitate partially or completely cross the volar ordorsal radial lines, respectively. These injuries can beclassified as midcarpal dislocations (Fig 21). Most

likely, lunate, midcarpal, and perilunate dislocationsare a spectrum of the same carpal injury.

Fractures associated with carpal dislocations areoften easier to identify on the postreduction view.Ideally, a postreduction view should be obtained be-fore a cast or splint is applied. CT can be helpful inclarifying more complex cases.

Carpal-Metacarpal Dislocation

Carpal-metacarpal dislocations are uncommon.However, every trauma hand radiograph should becarefully examined for this injury because it can bedifficult to identify and is often missed. The disloca-tion is most often dorsal (Fig 22), although volardislocation can also occur (Fig 23). The fifth meta-carpal is involved in most cases, either alone or incombination with other metacarpals. Fifth and/orfourth metacarpal-carpal dislocations are often asso-ciated with fractures of the hamate, typically anavulsion fracture off the dorsal lip of the hamate.

Figure 13. Floating elbow.Radiograph shows fracturesof humeral shaft and fore-arm.

Figure 14. Elbowfracture-dislocation.(a) AP and (b) lateralviews show fracturesof proximal ulna,coronoid process,and olecranon andposterior dislocationof radial head. Noteimpacted fracture ofradial head (arrow-head).

Hig

h-En

ergy In

juries to

the U

pp

er Extremity

183

Figure 15. Posterior elbow fracture-dislocation. (a) AP and (b) lateral views show posterior dislocation of radius and ulna with dis-placed radial head fracture (arrow). (c) Postreduction view shows that elbow joint dislocation has been reduced, and severely commi-nuted fracture of radial head (arrow) is well demonstrated.

Figure 16. Anterior fracture-dislocation of elbow. (a) Lateralview shows fracture through dis-tal olecranon, with anterior dislo-cation of radial head and ante-rior displacement of adjacentproximal ulna. (b) Postreductionlateral view shows that olecra-non fracture and radial headfracture have been nearly ana-tomically reduced. Radial headfracture (arrowhead) is partiallyobscured by coronoid process.

Figure 17. Monteggia fracture. (a) APand (b) lateral views show extraarticularfracture of the proximal ulna with anteriordislocation of radial head.

Gill

esp

y

184

Figure 18. Normal wrist.(a) On this properly positionedPA view, there are parallel lines(arrowheads) at second throughfifth metacarpal-carpal joints.(b) Lateral view. (c) Dia-grammed lateral view showsthat capitate articulates in fossaof lunate. Carpal bones shouldnot cross volar radial line (V) ordorsal radial line (D).

Figure 19. Perilunate dislocation. (a) PAview shows loss of normal congruency be-tween articular surfaces of lunate and capi-tate, and proximal carpal row is disrupted.Ulnar styloid fracture is also seen. (b) Capi-tate (arrowheads) is dislocated dorsally andcrosses dorsal radial line, while lunate isnormally aligned with distal radius.

Figure 20. Lunate dislocation. (a) PAview shows that lunate is displaced laterallyand is pie-shaped (arrowheads). There isalso fracture of radial styloid. (b) Lateralview shows that lunate (arrow) crossesvolar radial line, while capitate (arrow-heads) and remainder of carpus are nor-mally aligned with radius. Injury is classifiedas transradial lunate dislocation because ofradial styloid fracture.

Hig

h-En

ergy In

juries to

the U

pp

er Extremity

185

On a properly positioned PA view of the hand orwrist, the bases of the second through fifth metacar-pals form parallel lines with the adjacent carpalbones (Fig 18). Loss of these parallel lines suggestsmalalignment, especially if the metacarpal base andadjacent carpal bone overlap. Carpal-metacarpal dis-location should also be suspected whenever there isa fracture of a metacarpal base or adjacent carpalbone. However, the metacarpal-carpal joints are onlyclearly identified on the PA view if the palm is flatagainst the film cassette, and proper positioning isoften not possible in the severely injured patient.

The diagnosis of carpal-metacarpal dislocation isestablished on the lateral view, where one or more

metacarpal bases are not in proper alignment withthe carpus. The lateral view of the hand can be chal-lenging to the observer unfamiliar with this injury.The second through fifth metacarpal shafts are oftennot precisely parallel, so the analysis should focus onthe metacarpal-carpal region. CT is helpful in clarify-ing cases with unclear findings on the conventionalradiographs.

Axial Carpal Dislocation

Axial dislocation of the carpus is a rare injury indi-cated by widening between the carpal bones andtheir associated metacarpals in the distal carpal row,typically seen in severe crush or blast injuries. The

Figure 22. Carpal-metacarpal dislocation. (a) PA view shows that normal parallel lines at metacarpal-carpal joints are missing atfourth and fifth metacarpals. White band (arrowhead) at base of fifth metacarpal indicates overlap of metacarpal and hamate, anddislocation. (b) Oblique and (c) lateral views show dorsal dislocation (arrow) of fourth and fifth metacarpals. Small bone fragments(arrowhead) are probably small fractures from dorsal lip of hamate.

Figure 21. Midcarpal dislocation. (a) PAview shows that articulation of lunate (ar-rowheads) and capitate (arrow) is disrupted,and there is gross malalignment of proximalcarpal row. Lunate is displaced toward ulnarstyloid and has pie-shaped configuration.(b) Lateral view shows volar subluxation oflunate (arrowhead), which touches volar ra-dial line, and dorsal subluxation of capitate(arrow) and most of carpus, which touchdorsal radial line. In addition, small fracturesoff dorsal lip of distal radius or proximal poleof scaphoid are shown.

Gill

esp

y

186

fourth and fifth metacarpal-hamate joint is most ofteninvolved, and severe soft-tissue injuries are common.

Other Carpal Dislocations

A wide variety of carpal dislocations and fracture-dislocations have been described, in addition to thetypical patterns described in the previous paragraphs.Recently, we observed a case in which the lunate and

Figure 23. Carpal-metacarpal dislocation. (a) PA view shows normal parallel lines at metacarpal-carpal joints are missing at fifthmetacarpal-carpal joint. Note large overlap between the base of fifth metacarpal (arrowhead) and hamate, in addition to fracture atbase of fifth metacarpal. (b) Oblique and (c) lateral views show volar dislocation of fifth metacarpal (arrow). Fracture of distal thirdmetacarpal is also seen.

Figure 24. Carpal dislocation. (a) PA and(b) lateral views show that lunate (arrow)and proximal pole of scaphoid (arrowhead)have been ejected into volar compartmentof distal forearm. Gas in soft tissues indi-cates open injury. Dislocation is classifiedas transscaphoid scapholunate dislocation.

the proximal pole of the scaphoid were ejected intothe volar compartment of the distal forearm (Fig 24).

Suggested ReadingRogers LF, ed. Radiology of skeletal trauma. 3rd ed. Philadel-

phia, Pa: Churchill Livingstone, 2002; 593–929 (chap 15–18).Dandy DJ, Edwards DJ, eds. Essential orthopedics and trauma,

4th ed. Philadelphia, Pa: Churchill Livingstone, 2003; 179–230 (chap 12, 13).

187

Imaging Low-Energy UpperExtremity Injuries1

Kinetic energy is defined as mass multiplied by velocity, and no absolute physiologicthreshold exists demarcating high energy and low energy. Further, radiologic evalua-tions are not specifically guided by the magnitude of injury forces but will generallyfollow limitations imposed on available imaging techniques by patient- and injury-specific factors. That said, for the purposes of our discussion, low-energy trauma is de-fined as a result of (a) a fall from standing or sitting height or (b) a motor vehicle acci-dent at a speed of less than 15 mph (24 km/h). In this chapter, we describe injuries ofthe upper extremity that generally, but not exclusively, result from low-energy trauma.

INJURIES TO THE CLAVICLE

The clavicle functions as a bony support connecting the trunk and the arm (1,2). Theclavicle is S-shaped and is tubular in the proximal and medial aspect and flattened dis-tally. The clavicle has multiple muscle insertions and origins that may lead to typicaldislocations in some cases of fracture. The coracoclavicular ligaments bind the clavicleto the coracoid process, and the acromioclavicular ligaments bind the clavicle to theacromion.

Because of the specific anatomic position of the clavicle with respect to the trunk,radiographs will always show some overlap of the clavicle, such that one is not able toisolate the clavicle and get two perpendicular views. The basic protocol is an antero-posterior view and a 50° cephalic angulated view. Angling the beam to 50° cephalad,as suggested by Neer, produces more detailed information concerning the medial as-pect of the clavicle. In the absence of obvious dislocation or subluxation at the acro-mioclavicular joint, stress views with 10–15 lb (4.5–6.8 kg) of weight are importantfor detection or exclusion of ligamentous injuries. In any case, the injured arm musthang down unsupported by the side of the body to avoid understaging ligamentousinjury. Computed tomography (CT) may be helpful, especially for detection of subtlefractures of the sternoclavicular side of the clavicle.

Injuries to the clavicle are frequent. They are most frequent in childhood, and theclavicle is the most common fracture location during birth. In adults, the mechanismof trauma is a fall directly on the shoulder in 90% of the cases. In rare cases, a directblow may lead to injuries of the clavicle. Clavicular injuries are rare in cases of a fallon the outstretched hand.

Viktor M. Metz, MD, and Marcel O. Philipp, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 187–196.

1From the Department of Radiology, Medical University of Vienna, Waehringer-Guertel 18-20, A-1090 Vienna, Austria(e-mail: viktor.metz@univie.ac.at).

Met

z an

d P

hili

pp

188

Eighty percent of the fractures are located withinthe middle third of the clavicle, with a typical step-offat the fracture site, caused by the pull of the sterno-cleidomastoid muscle on the medial side and theweight of the arm on the lateral side (Fig 1). In 15%of the cases, fractures of the clavicle are located in theouter third, and as mentioned previously, stress viewsshould be considered for evaluation of otherwise occultcoraco- and acromioclavicular ligamentous injuries(Fig 2). In 5% of the cases, the fractures are locatedwithin the inner third.

DISLOCATIONS OF THE ACROMIOCLAVICULARJOINT

Dislocations of the acromioclavicular joint account for12% of all dislocations within the shoulder (2,3). Themechanism of injury may be (a) direct, by a fall on theshoulder, or (b) indirect, by a fall on the outstretchedhand. The joint space width of the normal acromio-clavicular articulation is between 5 and 8 mm (2 mmin the elderly). The coraco-clavicular interval distancenormally is between 11 and 13 mm. Compared withthe contralateral side, an asymmetry of 3–4 mm maybe normal. Asymmetry of the distance between the co-racoid process and the clavicle of more than 5 mm,however, strongly suggests ligamentous injury.

Radiography should be performed with a view inthe anteroposterior direction and a 15° cephalic view.To distinguish between type 1 and type 2 ligamentousinjuries, as mentioned previously, views should beobtained without and with 10–15 lb (4.5–6.8 kg) ofweight. In subtle or forensic cases, magnetic resonance(MR) imaging may be helpful.

On radiographs, acromioclavicular joint disloca-tions appear in sequence according to the severity ofthe force. In the initial stages, there is just a stretch orpartial tear of the acromioclavicular ligaments, and ra-diographs are (a) normal without and with weights(type 1) or (b) normal without weights and abnormalwith weights (type 2). If there is a complete tear of theacromioclavicular ligaments, the tear will lead to awidening of the acromioclavicular joint space on ra-diographs obtained without weights (type 3), with orwithout superior displacement of the clavicle (Fig 3).In severe cases, there is an additional tear of the cora-coclavicular ligaments and detachment of the deltoidmuscle and the trapezoid muscle. In addition, theremay be a fracture of the clavicle, the coracoid process,or the acromion.

FRACTURES OF THE PROXIMAL HUMERUS

Fractures of the proximal humerus (2,4) are one of themost common fractures in the elderly. The incidenceis three times higher in women than in men becauseof the differential presence of osteoporosis. There is

often a combination of proximal humerus fractureswith other fractures (eg, proximal femur, distal radius,or the cervical spine). The mechanism of injury is afall on the outstretched hand or a direct blow.

For radiographic examination, an anteroposteriorview in internal and external rotation should be ob-tained. During internal rotation, the lesser tuberosityprojects medially, and the globe-shaped humeral headprofiles the posterosuperior humeral head (eg, usualsite of the Hill-Sachs deformity). In external rotation,the greater tuberosity is projected more laterally, facili-tating its evaluation. A Grashey view (45° posterioroblique projection), which is a 40°–45° anteroposte-rior oblique view, should be obtained to profile theglenohumeral joint. An axillary lateral projection anda scapula Y view (45° anterior oblique projection)may be obtained for better evaluation of displace-ment. CT is of importance in complex comminutedfracture-dislocations.

As a classification scheme, the Neer classification isthe most widely used and is based on the indirect as-sessment of the functional integrity of the soft-tissuestructures supporting physiologic function (eg, peri-osteum, muscles of the rotator cuff). For develop-mental and anatomic reasons, Neer divided theproximal humerus into four parts: the articular head,the surgical neck, the greater tuberosity, and the

Figure 2. Radiograph showing fracture of clavicle (arrow) atouter third near acromioclavicular joint.

Figure 1. Fracture of middle third of clavicle. Anteroposteriorcephalic angulated view shows typical step-off at fracture site(arrows), caused by pull of sternocleidomastoid muscle on me-dial side and weight of arm on lateral side.

Imag

ing

Low

-Energ

y Up

per Extrem

ity Inju

ries

189

lesser tuberosity. A fracture of the humeral head isdefined as displaced if there is a displacement ofmore than 1 cm and/or if there is an angulation ofmore than 45° of any of the four parts mentionedpreviously. Grading of fractures of the proximal hu-merus is based on the total number of fracture partsthat are displaced: Grades range from a one-part frac-ture, representing any combination of nondisplacedfractures, to a four-part fracture caused by displace-ments of the fractures of the surgical and anatomicnecks and both tuberosities.

Isolated fractures are rare. There are often combina-tions of fractures, and the number of complicationsincreases with the number of fractured parts. In about80% of the cases, there is no displacement, and thefracture is classified as a one-part fracture, which isconsidered stable. Because of the muscles that coverthe humeral head, fractures commonly have an egg-shell appearance (Fig 4). Often there is an impactionof the surgical neck. Fractures at the anatomic neck ofthe humeral head may disrupt blood supply and leadto osteonecrosis of the humeral head.

Complications of proximal humerus fractures areinjuries of the brachial plexus and/or the axillary ar-tery. Those complications are most likely in anteriorlydisplaced fractures of the surgical neck. Therefore, youshould direct careful clinical examination to the possi-bility of those complications. Duplex ultrasonographyor CT angiography may be helpful in predicting andgrading vascular injuries. Another complication is theso-called frozen shoulder, which is caused by fibrosisand adhesive capsulitis. This complication may beprevented by early active or passive mobilization.

GLENOHUMERAL JOINT INJURIES

The glenohumeral joint consists of the shallow gle-noid augmented by the cartilaginous labrum and sur-rounded by numerous muscles, which provides thegreatest prehensile motion (2,5). For conventional ra-diography, the anteroposterior view, the Grasheyview, the axillary lateral view, and the scapula Y viewmay be obtained. After reduction of anterior disloca-tions, an apical oblique view (Grashey projection with

Figure 3. Acromioclavicular dislocation. (a) Radiograph shows that joint space on right is widened, and there is step-off comparedwith left side (arrows). (b) Coronal MR image shows that joint space (star) is obviously widened. Coracoclavicular ligaments are tornand retracted (arrow).

Figure 4. Two-part fracture ofhumeral head at site of greatertuberosity and site of surgicalneck. (a) Radiograph showstypical eggshell appearance (ar-row) because of coverage ofmuscles. (b) In external rotation,fracture of greater tuberosity(arrow) and fracture of surgicalneck (star) are more evident onradiograph. Note presence ofosteoporosis.

Met

z an

d P

hili

pp

190

an additional 45° caudal angulation of the central rayof the x-ray beam) best shows both the Hill-Sachs andbony Bankart deformities. For detection or exclusionof associated injuries, CT and MR imaging are impor-tant adjunctive modalities.

Glenohumeral dislocations are among the mostcommon dislocations within the body. In as many as95% of the cases, the dislocation is anterior because ofexternal rotation and abduction or because of indirecttrauma (Fig 5). In those circumstances, the posterolat-eral surface of the humeral head impacts against theanteroinferior surface of the glenoid fossa. A lesscommon trauma mechanism is a blow to the postero-lateral side of an abducted arm, leading to a displace-ment of the humeral head into the area of the sub-coracoid fossa. Posterior dislocations are rare andtypically are subsequent to seizures, electrocution,or a posterior force with the arm in internal rotation.Superior dislocations, a rare variant of posterior dis-locations, are thought to be due to a blow on theflexed elbow with the arm abducted. Luxatio erecta,an uncommon anterior dislocation variant, occurswith severe hyperabduction of the arm where the neckof the humerus impinges against the acromion, driv-ing the head of the humerus both distal and anterior.

The anterior glenohumeral dislocation with dis-placement of the humeral head into the subcoracoid

recess may lead to an impaction fracture of the poste-rolateral aspect of the humeral head against the ante-rior aspect of the glenoid and is known as a Hill-Sachsdeformity (Fig 6). Similarly, anterior glenohumeraldislocation may cause a cartilaginous or osseocarti-laginous fracture at the anteroinferior aspect of theglenoid, which has been described as a Bankart lesionor a bony Bankart lesion, respectively (Fig 7). Insubtle cases, CT and MR imaging may be helpful fordepiction of Hill-Sachs or bony Bankart lesions. Inaddition, MR imaging may be performed for furtherevaluation of other associated injuries, such as liga-mentous tears or injuries of the rotator cuff. MR ar-thrography is the method of choice for precise evalua-tion of injuries of the cartilaginous labrum (Fig 8).

Although posterior glenohumeral dislocations arerare, it should be mentioned that these injuries areclinically and radiographically often occult. Posteriorglenohumeral dislocations are missed in 60% of thecases at initial evaluation. Posterior glenohumeral dis-locations clinically appear as the inability to rotate thearm externally. On conventional radiographs, a di-rected search for the rim sign and the trough sign (Fig9) will help to avoid misinterpretation of posteriorglenohumeral dislocations.

Figure 5.Anterior gle-nohumeraldislocation(arrow), withdisplacementof the humer-al head inarea of cora-coid process,on (a) antero-posterior and(b) axillarylateral views.(a) Note im-paction frac-ture of hu-meral head(arrowhead),which is de-scribed asHill-Sachsdeformity.

Figure 6.(a) Anteriordislocation ofglenohumeraljoint on radio-graph.(b) Axial CTimage clearlyshows Hill-Sachs defor-mity (arrow).

Imag

ing

Low

-Energ

y Up

per Extrem

ity Inju

ries

191

INJURIES OF THE ELBOW

The elbow consists of three joints: the humeroulnarjoint, the humeroradial joint, and the radioulnar joint(6,7). For conventional radiography, the standard viewsare (a) an anteroposterior view in the supine positionwith full extension of the elbow and (b) a lateral viewwith the elbow flexed 90° with the lower arm and withthe hands in the neutral thumb-side-up position. Addi-tional views include internal and external oblique viewsand the radio-capitellum view (positioned as for a neu-tral lateral view and additional angling of the central rayof the x-ray beam 45° cephalad). In complex or subtlecases, CT or MR imaging will lead to precise diagnosis.

Radial head fractures are the most common elbowinjuries in adults and account for approximately 60%of all elbow injuries. The mechanisms of trauma in-clude a fall on the outstretched hand or a force trans-mitted to the long axis of the radius, leading to a com-pression of the radial head against the capitellum. TheMason classification is the most commonly used clas-sification scheme for radial head fractures. The Masontype I fracture is a fracture with a displacement of thefracture fragment of less than 2 mm (50% of thecases) (Fig 10). Type II fractures are segmental frac-

tures consisting of one-quarter of the radial head witha displacement or compression of more than 2 mm,and type III fractures are defined as comminuted frac-tures of the entire radial head.

Fractures of the olecranon account for 20% of elbowfractures. The mechanisms of trauma are a fall on theelbow with the elbow flexed or a direct blow to the el-bow. In most cases, the fracture line runs transverselyand may be overlooked on the anteroposterior view(Fig 11) but commonly is easily diagnosed on the lat-eral view. In rare cases, the fracture line runs obliquelybecause of valgus forces.

Coronoid process fractures are rarely isolated frac-tures and are most commonly associated with posteriordislocation of the elbow (Fig 12). Isolated fractures ofthe coronoid process may be difficult to see on antero-posterior and lateral views. Therefore, oblique viewsshould be obtained.

FRACTURES OF THE SHAFT OF THE RADIUSAND/OR ULNA

These fractures are most commonly caused by a fall onthe outstretched hand or a longitudinal compression

Figure 9.Posterior gle-nohumeraldislocation onradiograph.Joint space ismore than 6mm wide(white arrow),which is de-scribed as rimsign. Note im-paction frac-ture (black ar-rows) seen asvertical line.This has beendescribed astrough line.

Figure 7. Bony Bankart lesion(arrow) after anterior dislocationof glenohumeral joint on (a) axialCT image and (b) coronal CTreformation.

Figure 8.Fat-sup-pressed axialMR image ofshoulder afteranterior dislo-cation. Thereis clear evi-dence of car-tilaginousBankart le-sion (arrow)at anterioraspect of gle-noid. In addi-tion, note areaof high signalintensity (star)at humeral head, indicating Hill-Sachs impression fracture.

Met

z an

d P

hili

pp

192

force (6,8). In most cases, both the radius and the ulnarshaft are fractured. Isolated single-bone fractures areless common. Not uncommonly, one of the bones isfractured while the other is dislocated, and usually thefracture shows obvious displacement or rotation.

The Monteggia fracture is defined as a fracture of theulna with a dislocation of the proximal radius. Thetrauma mechanism is a fall on the outstretched handcombined with a forced pronation of the forearm. Theradiocapitellar line (created by bisecting the radial shaftproximal to the radial tubercle) should always bedrawn to avoid overlooking subluxation or dislocationof the proximal radius. Under normal conditions, theline runs through the central third of the capitellum.

The Galeazzi fracture (Fig 13) is an isolated frac-ture of the distal radius in combination with sublux-ation at the distal radioulnar joint. In contrast toMonteggia fractures, the mechanism of trauma inGaleazzi fracture-dislocations is a fall on the out-stretched hand with maximal pronation of the arm,although a direct blow may uncommonly lead tothe injury. Subluxation or dislocation of the distalradioulnar joint may be overlooked. Therefore, aneutral lateral view of the distal forearm, which canbe achieved by adducting the arm to the trunk andflexing the elbow 90° and the hand 90° to the cas-sette (in between maximal pronation and supina-tion), should be searched for distal, dorsal, or ulnar

Figure 10. Fracture of radialhead. (a) On anteroposteriorview, there is no clear evidenceof fracture. (b) On lateral viewwith elbow flexed 90°, note frac-ture of radial head (arrow) withdisplacement of more than 2 mm(Mason type II).

Figure 11. Fracture of olecra-non (arrow) on (a) anteroposte-rior and (b) lateral views.

Imag

ing

Low

-Energ

y Up

per Extrem

ity Inju

ries

193

displacement of the distal ulna relative to the distalradius metaphysis.

DISTAL RADIUS FRACTURES

Fractures of the distal radius are the most common frac-tures of the skeleton (8,9). The mechanism of trauma ismost commonly a fall on the outstretched hand, al-though less commonly the fracture may occur becauseof high-energy trauma. Fractures of the distal radius arecommon in children, because of physeal injuries, andare also common in the elderly, because of osteoporo-

sis. In the elderly, concurrent fracture of the distal radiusand fracture of the humeral head or the cervical spinemay occur with falls from standing or seated height.

Several classification schemes for distal forearmfractures have been introduced, but none is widely ac-cepted. For example, Fryckman has introduced a clas-sification scheme for distal radius fractures that recog-nizes the importance of associated ulna fractures, andhe has classified distal radius fractures into eighttypes (Fig 14).

The most common type of distal radial fracture isthe Colles fracture, which occurs because of a fall on

Figure 12. Fracture of coronoid process associated with posterior dislocation of elbow. (a) On lateral radiograph, there is evidenceof fracture of radial head (arrowhead) and dislocation of elbow (arrow) but no evidence of fracture of coronoid process. (b) On antero-posterior view, fracture of coronoid process is not well shown (lower arrow), but there is clear evidence of dislocation of joint (upperarrow). (c) On CT reformation, fractures of radial head (short arrow) and coronoid process (arrowhead), with displacement of smallfracture fragment (long arrow), are well depicted.

Figure 13. Galeazzi fracture. (a) On pos-teroanterior radiograph, fracture of distal ra-dius shaft (long arrow) and some sublux-ation of distal radioulnar joint (short arrow)can be seen. (b) On lateral radiograph, dis-location of distal radioulnar joint (arrow) ismuch more evident.

Met

z an

d P

hili

pp

194

the dorsally flexed hand. In this Colles fracture, thedistal fracture fragment is displaced dorsally (Fig15). The classic Colles fracture is an extraarticularfracture. The second most common fracture of thedistal radius is the Smith fracture, which, in contrastto the Colles fracture, occurs because of a fall on thepalmar flexed hand. In this type of fracture, the dis-tal fracture fragment is displaced or angulated pal-mar. The typical Smith fracture is also an extra-articular fracture.

Barton has described a third type of distal radiusfracture. Barton fractures and reverse Barton frac-tures are intraarticular fractures of the anterior orposterior rim of the distal radius. These types offractures occur after a more axially applied force.

The chauffeur fracture is an intraarticular fractureof the radial styloid process. This type of fracturearises after an axially applied force or is caused by adirect blow (Fig 16).

In any type of distal radius fracture, a careful ex-amination of the carpal bones should be performedbecause distal radial fractures are often associatedwith scaphoid fractures or intercarpal ligamentousinjuries (eg, scapholunate ligament injuries). Al-though there are numerous classifications and clas-sification schemes for distal radial fractures, the im-portant question is what the surgeon or orthopedistwants to know. Therefore radial inclination and pal-mar inclination should be reported. In normal indi-viduals, average radial inclination is 21°, and meanpalmar inclination is 11°. In addition, the radiolo-gist should report whether the fracture is commi-nuted and whether the fracture is intra- or extra-articular. In addition, bone quality (eg, osteoporo-sis) is important to know for further treatment andshould be mentioned.

DISLOCATIONS OF THE DISTAL RADIOULNARJOINT

Dislocation of the distal radioulnar joint may occur asan isolated injury (9). More commonly, dislocationsof the distal radioulnar joint are found in associationwith distal radius fractures. For radiologic diagnosis ofdislocation of the distal radioulnar joint, it is impor-tant to obtain a true neutral lateral radiograph. Thetrue lateral radiograph is obtained by adducting thearm to the trunk and flexing the elbow 90°, and thedistal forearm forms a straight line with the metacar-pal bones and is perpendicular to the cassette. Evenslight supination or pronation must be avoided. Withthis arrangement, the distal radius and the distal ulnaare superimposed.

FRACTURES OF CARPAL BONES

The wrist consists of complex bony and ligamentousanatomic structures (9,10). Standard radiographsshould include a posteroanterior view in the neutralposition (the arm abducted 90°, the elbow flexed 90°,

Figure 15. (a) Posteroanteriorand (b) lateral views of patientwith Colles fracture. Typically,distal fracture fragment is dis-placed and/or angulated dorsally(arrow).

Figure 14. Schematic of Fryckman classification of distal ra-dius fractures.

Imag

ing

Low

-Energ

y Up

per Extrem

ity Inju

ries

195

Figure 18. Arcsdescribed by Gilula.On normal wrist ra-diographs, threesmooth arcs can bedrawn along prox-imal and distal car-pal rows. Arc I joinsconvexity of proxi-mal carpal row, arcII joins concavity ofproximal carpalrow, and arc IIIjoins convexity ofthe distal carpalrow. In somepeople, there maybe small step-off atsite of the lunotri-quetral joint, whichis normal.

Figure 16. Chauffeur fracture(arrow) without dislocation on(a) posteroanterior and (b) lat-eral radiographs.

Figure 17. Vul-nerable zone forcarpal bone frac-tures. Lunate boneis not included invulnerable zone.

and the hand flat on the cassette, with the radius andthe third finger colinear), a 30° pronated oblique view,a true neutral lateral view, and a dedicated scaphoidview (eg, a posteroanterior view with ulnar deviationof the wrist and 20° proximal angulation of the cen-tral ray of the x-ray beam). CT is helpful in complexcases. MR imaging is important in patients with ques-tionable occult fractures and for precise evaluation ofcarpal ligaments and the triangular fibrocartilage com-plex. The mechanism of trauma for carpal fracturesmost commonly is a fall on the outstretched handwith a slight dorsiflexion and ulnar deviation andsome type of supination.

A vulnerable zone (Fig 17) has been described forcarpal bone fractures. This zone includes the scaphoid,the trapezium, the trapezoid, the capitate, and the tri-quetrum and excludes the lunate. Gilula has describedthree arcs (Fig 18), which are smooth parallel linesthat should be evaluated carefully to assist in the rec-ognition of intercarpal malalignments. A broken arcor a step-off to any side of one or more arcs alwaysindicates fracture, dislocation, or subluxation.

SCAPHOID FRACTURES

Scaphoid fractures constitute 60%–70% of all carpalbone fractures (9,10). They are common in the 15- to40-year-old age group and are rare in children and inthe elderly. Seventy percent of the fractures of thescaphoid run through the scaphoid waist, 20% runthrough the proximal pole, and 10% run through thedistal pole. Avascular necrosis of the proximal pole ofthe fractured scaphoid may complicate scaphoid frac-tures and may result in activity-limiting loss of wristand hand function. Fractures at greatest risk of avascu-lar necrosis include those displaced greater than 2

Met

z an

d P

hili

pp

196

mm, and the likelihood of avascular necrosis directlyincreases by scaphoid waist fracture location from dis-tally to proximally. In the distal fifth of the scaphoid,necrosis occurs in 15% of the cases, and in the proxi-mal fifth of the scaphoid bone, necrosis occurs innearly 100% of the cases (Fig 19). MR imaging is animportant imaging modality for early detection of thiscomplication because MR imaging allows immediatediagnosis and classification of avascular necrosis.Only early stages of avascular necrosis (grades 1 or 2)may respond to therapy. If avascular necrosis of thefractured scaphoid bone is depicted on conventionalradiographs, primary treatment of the fracture maynot be possible. Other complications of scaphoid frac-tures include nonunions or delayed unions. CT, aswell as MR imaging, may be helpful in the diagnosisof these potential complications.

References1. Craige EV. Fractures of the clavicle. In: Rockwood CA,

Green DP, eds. Fractures in adults. 4th ed. Vol 1. Philadel-phia, Pa: Lippincott-Raven, 1996; 1109–1161.

2. Rogers LF. The shoulder and humeral shaft. In: Rogers LF,ed. Radiology of skeletal trauma. 2nd ed. Vol 1. New York,NY: Churchill Livingstone, 1992; 653–748.

3. Rockwood CA Jr, Williams GR, Young DC. Injuries to theacromio-clavicular joint. In: Rockwood CA, Green DP, eds.Fractures in adults. 4th ed. Vol 2. Philadelphia, Pa: Lippin-cott-Raven, 1996; 1341–1413.

4. Bigliani LU, Flatow EL, Pollock RG. Fractures of the proxi-mal humerus. In: Rockwood CA, Green DP, eds. Fracturesin adults. 4th ed. Vol 1. Philadelphia, Pa: Lippincott-Raven,1996; 1055–1107.

5. Rockwood CA Jr, Wirth MA. Subluxations and dislocationsabout the glenohumeral joint. In: Rockwood CA, Green DP,eds. Fractures in adults. 4th ed. Vol 2. Philadelphia, Pa:Lippincott-Raven, 1996; 1193–1339.

6. Rogers LF. The elbow and forearm. In: Rogers LF, ed. Ra-diology of skeletal trauma. 2nd ed. Vol 2. New York, NY:Churchill Livingstone, 1992; 749–836.

7. Hotchkiss RN. Fractures and dislocations of the elbow. In:Rockwood CA, Green DP, eds. Fractures in adults. 4th ed.Vol 1. Philadelphia, Pa: Lippincott-Raven, 1996; 929–1024.

8. Richards RR, Corley FG Jr. Fractures of the shaft of the ra-dius and ulna. In: Rockwood CA, Green DP, eds. Fracturesin adults. 4th ed. Vol 1. Philadelphia, Pa: Lippincott-Raven,1996; 869–928.

9. Rogers LF. The wrist. In: Rogers LF, ed. Radiology of skel-etal trauma. 2nd ed. Vol 2. New York, NY: ChurchillLivingstone, 1992; 837–938.

10. Cooney WP III, Linscheid RL, Dobyns JH. Fractures anddislocations of the wrist. In: Rockwood CA, Green DP, eds.Fractures in adults. 4th ed. Vol 1. Philadelphia, Pa: Lippin-cott-Raven, 1996; 745–867.

Figure 19. (a) Posteroanterior radiographwith fracture (arrow) at proximal fifth ofscaphoid. (b) Corresponding coronal T1-weighted MR image demonstrates earlystage of avascular necrosis of proximalfracture fragment (arrow), while distal partof scaphoid bone (star) is vital.

197

Low-Energy Injuries ofthe Lower Limb1

The distinction between low- and high-energy injuries is arbitrary, but for the purposesof this chapter, the low-energy section will consider injuries sustained as a result offorces applied in a repetitive or sporting setting and will also include falls and contu-sions sustained through direct impact. The injuries will be considered according to bodypart, in the following order: (a) pelvis and hips, (b) thigh, (c) knee, (d) lower part of theleg, and (e) ankle and foot; but muscle injury will be considered initially to establishspecific principles and patterns that will subsequently be qualified by examples of inju-ries commonly encountered in each of the specified areas.

MUSCLE INJURY

Muscle injury is common and is mainly encountered in the setting of sports-relatedactivity. There are two main types of injury, each of which has a typical clinical historyand imaging appearance. The more common type is the muscle strain or tear, usuallysustained as a result of isometric muscle contraction, while the muscle contusion oc-curs through direct impact during combative sports or consequent to a fall.

Muscle Sprains and Tears

Tears and sprains are some of the most common injuries and are particularly preva-lent in the lower limb. Muscles particularly prone to injury during isometric contrac-tion include the so-called two-joint muscles that bridge two articulations, includingthe hamstring muscles, rectus femoris, and gastrocnemius. Although the exact muscle-tendon configuration may vary, the basic principle dictates that the tendon extendsinto the muscle and that the site of maximum weakness that is most prone to injury isthe muscle-tendon junction, where elasticity is limited.

There are three basic grades of injury. Although an element of muscle or muscle-tendon injury will be microscopically evident in all grades, ultrasonography (US) andmagnetic resonance (MR) imaging show no macroscopic fiber disruption. Grade 1injuries usually show a localized increase in echogenicity at US (Fig 1), although dif-fuse hypoechoic areas may also be observed. Increased edema associated with grade 1injuries is manifest as diffuse high signal intensity on short inversion time inversion-recovery (STIR) or equivalent T2-weighted fast spin-echo (SE) MR images obtainedwith fat saturation (Fig 2) (1).

Philip M. Hughes, MBBS, MRCP, FRCP

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 197–215.

1From X-ray West, Derriford Hospital, Derriford Rd, Plymouth, Devon, England PL6 8DH (e-mail: philip.hughes@phnt.swest.nhs.uk).

Hu

gh

es

198

Grade 2 injuries represent overt muscle tears, andclinically the muscle has a restricted range of move-ment and is tender to palpation. US reveals focal fiberdisruption that appears hypoechoic and is usually iden-tified at the muscle-tendon junction. The disruption atthese sites may vary from a minor tear to near total rup-ture (Figs 3, 4). MR imaging is equally efficacious, withthe proviso that muscle edema can sometimes makethe distinction between grade 1 and minor grade 2 in-juries difficult (Fig 5). Grade 1 and grade 2 injuries areusually more easily distinguished at US. The clinical ef-fect can be important because grade 1 injuries usuallyallow a return to full activity in 10–14 days, whilegrade 2 injuries may take as long as 4 weeks.

Grade 3 injuries represent complete muscle rup-ture, a condition that may be accompanied by con-siderable discomfort. Not uncommonly, however,symptoms may be relatively mild in individuals whohave completely disconnected the muscle unit, butfunction will be absent. These patients may requiresurgical repair, but most are managed conservatively.Recovery depends on whether healing is feasible. If itis, 3 months of reduced activity is the norm. If thetear is distal, the tendon end may be retracted andnot likely to heal. However, if other muscle groupsare compensating, some patients return to sportingactivity within 2–3 weeks (Fig 6).

The grades of these injuries are rough guides to therecommended recuperative period, which is furtherinfluenced by recurrent injury and uncorrected bio-mechanical imbalances, which may require a longerrecovery phase. Grade 2 injuries vary greatly from mi-nor tears, requiring 4 weeks of rehabilitation, to severenear-total tears, requiring 12 weeks. The rehabilitationperiod is also inversely related to the amount of expertphysiotherapy and supervised retraining.

Muscle Contusion

Muscle contusion occurs following direct impact,results in muscle fiber disruption, rather than muscletendon injury, and is often accompanied by intra-muscular hemorrhage and hematoma formation.The muscle is swollen and acutely tender, with severelimitation in the range of movement. Compartmentsyndrome can ensue, and in the presence of severepain or neurovascular compromise, US-guided orsurgical hematoma evacuation may be required.

Hematomas associated with contusion may haveill-defined irregular margins when associated with fi-ber disruption, but most are well defined and oval in

Figure 1. Grade 1 hamstring injury in a 24-year-old soccerplayer. US images of right (R) and (L) hamstring compartmentsin transverse section show increased echogenicity in region ofinjured right semimembranosus muscle but no fiber disruption.

Figure 2. Grade 1 hamstring injury in a 44-year-old elite dis-tance walker. Transverse STIR MR image demonstrates subtleincreased signal intensity in injured semimembranosus muscle(arrows) with prominent vessels (arrowhead) indicative of hy-peremia.

Figure 3. Grade2 tear of gastroc-nemius in a 34-year-old runner.Sagittal US imagedemonstrates fo-cal hypoechoiczone with loss offiber continuity atmedial side ofgastrocnemiusmuscle-tendonjunction (arrow),indicative of local-ized tear.

Figure 4. Massin anterior midpor-tion of thigh of a15-year-old soccerplayer. Sagittal USimage with quadri-ceps contractedshows proximal re-traction and masseffect of torn rectusfemoris muscle (ar-rows), which otherimages demon-strated to be a partial tear.

Low

-Energ

y Inju

ries of th

e Low

er Limb

199

configuration. Within days of injury, acute hemato-mas show high-signal-intensity areas on T1-weightedSE MR images because of the presence of methemo-

Figure 5. Partial grade 2 tearof rectus femoris in a 21-year-old female high-hurdler. (a) Sag-ittal STIR MR image demon-strates focal high signal inten-sity with fluid characteristics in-dicative of rectus femoris tear.(b) Transverse STIR MR imageshows residual intact musclecomponent medially (arrow).

Figure 6. Mildtenderness andbruising along pos-terolateral portionof knee, with bicepsfemoris tear, in a28-year-old soccerplayer. Patient re-turned to profes-sional soccer within4 weeks. (a) Sagit-tal far lateral STIRMR image showsrupture and retrac-tion of biceps femo-ris tendon (arrow).(b) Equivalent sag-ittal US image dem-onstrates "bell-clapper" configura-tion of retractedtendon end (arrow)and surroundinghematoma.

Figure 7. Persis-tent mass in ante-rior part of thigh fol-lowing contusion ina 44-year-old man,despite a delayedsurgical drainage.US image showschronic hypoechoichematoma with re-tained drain (ar-row), which wasunsuspected attime of referral. Nosepsis was identi-fied when this he-matoma was sub-sequently drainedsurgically.

globin. With increasing age, hematomas acquire amore homogeneous fluid density and appear hypo-echoic on US images (Fig 7), low in signal intensityon T1-weighted SE MR images, and high in signal in-tensity on T2-weighted SE or STIR MR images. Calcifi-cation is commonly associated with muscle contu-sion (Fig 8). Hematomas are usually avascular, butrarely, organizing hematomas or areas of myositisossificans may manifest as hyperemic masses. Con-tused muscle in the reparative hyperemic phase isprone to acute hemorrhage if the muscle unit is inad-vertently loaded and the neovascularization disrupted.

Sequelae of Muscle Injury

There are several late sequelae of muscle injury.Masses may manifest in the lower limb, representingeither (a) retracting muscle fibers caused by a substan-tial tear with no healing or (b) muscle herniation.

Muscle hernias.—Muscle hernias and retractilemuscle masses frequently involve the rectus femorisand are more obvious with the muscle functioning.The examination should always be performed with thepatient standing or while isometrically contracting themuscle unit. Hernias are due to trauma-related de-fects in the epimysium investing the muscle (Fig 9),

Hu

gh

es

200

and retractile muscle masses are indicative of sub-stantial partial or complete rupture of the muscleunit with retraction of the ends of the disruptedmuscle (Fig 10). Neither of these abnormalities canbe positively diagnosed with MR imaging but ben-efit from dynamic US evaluation.

Fibrous scarring.—Fibrous scarring results from se-vere or recurrent injuries, but most muscle injuriesheal without cicatrization. Fibrosis is demonstrable atMR imaging as an area of low signal intensity with allpulse sequences (Fig 11) and is seen at US as a linearor stellate hyperechoic area (Fig 12). The fibrosis af-fects the muscle-tendon junction and predisposes torecurrent tears.

Hematomas.—Hematomas are more commonlyassociated with contusion, rather than tears. He-matomas may manifest as a mass lesion that can bedifficult to distinguish from a tumor without theappropriate medical history, which may on occa-sion require focused questioning of the patient fol-lowing imaging. This is particularly true of sports-men or children because the history may not be im-mediately associated with the mass in the patient’smind if there is a temporal delay in manifestation.Conventional radiographs and computed tomogra-phy (CT) show curvilinear peripheral calcificationin a long-established mass, a feature characteristicof hematoma and atypical for mineralizing sarco-mas (Fig 13).

Myositis ossificans.—Myositis ossificans follows severecontusions and reportedly occurs in 9%–20% of suchinjuries, particularly in the anterior thigh and adductorcompartments (2–4). When superficial muscle groupsare affected, patients present with (a) severe pain, oftendisproportionate to the initial injury, (b) reducedfunction, and (c) a mass. The mass has ill-definedmargins at US and MR imaging, enhances in the earlystages after administration of gadolinium-based con-trast material, and shows neovascularization at US.Differentiation from a soft-tissue tumor can be diffi-cult without the characteristic history of severe local-ized trauma.

Myositis ossificans is distinct from osseous avul-sion. The calcification is most often sheetlike, is ori-

entated in the long axis of the injured muscle, andemanates from pluripotential mesenchymal musclecells. Less frequently, a more discrete mass emanatingfrom and attached to bone may be encountered. Thelatter may represent rupture of subperiosteal hemor-rhage into the surrounding tissues. The associated cal-cification is first identified on conventional radio-graphs approximately 6 weeks following injury. US isa highly sensitive method of identifying myositisossificans (Fig 14) as early as the 3rd week and is gen-erally more sensitive at identifying calcification than isMR imaging (Fig 15) (5).

Calcific myonecrosis.—Whereas myositis ossificans isassociated with a relatively recent history of trauma,calcific myonecrosis is a late sequel, occurring be-tween 10 and 64 years after the initial episode oftrauma (6). Trauma is a common antecedent of cal-cific myonecrosis, like myositis ossificans. However,calcific myonecrosis may also follow thermal, elec-trical, or neurologic injury, where a compartment

Figure 8. Calcification in a 24-year-old rugby player with re-current quadriceps pain after a contusional injury that had oc-curred months earlier. Transverse US images of both thighswere obtained. Hyperechoic foci with posterior shadowing inright rectus femoris muscle represents dystrophic calcification(arrows).

Figure 9. Intermittent lump (muscle her-nia) on anterolateral side of lower portionof right leg of a 15-year-old soccer player.(a) Sagittal US image of normal (left) tibialisanterior muscle shows epimysium (arrows).(b) Sagittal US image of symptomatic rightside obtained with patient standing showsloss of definition of investing epimysium,with bulging of tibialis anterior muscle atsite of defect (arrows).

Low

-Energ

y Inju

ries of th

e Low

er Limb

201

Figure 12.Recurrenttears alongposterome-dial side oflower portionof leg of a44-year-oldjogger. Sagit-tal US imageshows linearhyperechoiczone (arrows)within distalgastrocne-mius adjacentto muscle-tendon junction, indicative of scarring.

Figure 11.Recurrentquadricepsinjury in a35-year-oldathlete.TransverseSTIR MR im-age demon-strates tearof rectusfemoris (shortarrow), withlow-signal-in-tensity fibro-sis (arrowhead) surrounding normal central tendon (long arrow).

Figure 10. Intermittent mass inanterior portion of thigh of a 25-year-old marine. Sagittal US im-ages of rectus femoris muscle inrelaxed and contracted statesconfirm substantial muscle tear,with mass representing the re-tracting proximal muscle fibers(arrows).

Figure 13. Mass in left groinof a 12-year-old boy. History oftrauma was eventually elicited.(a) Pelvic radiograph and(b) transverse CT image showperipheral well-defined curvilin-ear calcification indicative of os-sifying hematoma. Findings fromother imaging examinationswere less diagnostic: (c) Radio-isotope bone scan after adminis-tration of technetium-labeledmethylene diphosphonate (MDP)demonstrates nonspecific areaof increased uptake ("hot spot").(d) Transverse T2-weighted fastSE MR image demonstratescentral mass lesion with sur-rounding edema.

syndrome and subsequent myonecrosis of themuscle unit represent the single common pathwayshared by these varied causes. Patients present withan expanding mass, commonly in the lower portionof the leg, and show platelike calcification extendingthroughout at least one muscle compartment (Fig 16).

Hu

gh

es

202

Figure 14. Myositis ossificansin a 27-year-old professionalsoccer goalkeeper with injury toleft thigh. Injury was sustained 8weeks prior to imaging. (a) Coro-nal US image of asymptomaticthigh demonstrates linear inter-face between femur (arrows)and adductor magnus muscle.(b) US image of injured thighshows undulating irregularechogenic mass at muscle-tendon junction of adductormagnus, which represents myo-sitis ossificans (arrows).

Figure 15. Myositis ossificans in right thigh of a 10-year-old boy with injury sustainedin bicycling accident. (a) Conventional radiograph and (b) transverse CT image showsheetlike ossification indicative of myositis ossificans. (c) Nonenhanced and (d) gado-linium-enhanced sagittal T1-weighted SE MR images show rim enhancement in peri-osseous region along most of length of femur, representing resolving hematoma, butmyositis ossificans cannot be appreciated.

The calcification is amorphous, unlike that of myo-sitis ossificans, which exhibits a trabecular structurewhen mature. Because calcific myonecrosis occurs inan older age group and occasionally involves theperiosteum, the condition must be distinguishedfrom neoplastic masses. The extent and the compart-mentalization of calcification are indicative of thediagnosis. MR imaging shows a partially cystic massand no evidence of soft-tissue enhancement (7).

Low

-Energ

y Inju

ries of th

e Low

er Limb

203

PELVIS AND HIPS

Low-energy injuries to the pelvic ring are encountered(a) in the form of stress fractures, when excessive forceis applied to the normal bones of the pelvis, or (b) asan insufficiency fracture, when normal forces appliedacross the pelvis result in fractures because of intrinsicdeficiency in bone structure (most frequently the re-sult of osteoporosis). A third distinctive pattern ofrelatively low-energy pelvic fractures includes avulsioninjuries, which are encountered predominantly in in-dividuals following sporting activity and are more fre-quent in the immature skeleton.

Stress Fractures

Stress fractures are sustained when the pelvis is sub-jected to excessive loading. This most frequently occursduring athletic training, in either elite or recreationalathletes. The vogue for distance running in the generalpopulation has caused an increase in stress injuries ofthe lower limbs, including the pelvis. Conventional ra-diographs usually show normal findings, and in thesecircumstances, MR imaging is the preferred method ofinvestigation. The injury is usually associated with mar-row edema, which may have a linear orientation and

occur anywhere in the pelvic ring. The posterior columnof the acetabulum and the sacrum (Fig 17) are favoredsites because of their relatively high loading (8). Sacralfractures or stress reactions often parallel the sacroiliacjoint. The fracture line may be identifiable but is oftenobscured on STIR or fat-saturated MR images and is bestshown on T1-weighted SE (Fig 18), T2-weighted fast SE(Fig 19), or gadolinium-enhanced MR images. The ab-sence of a fracture line is not uncommon and is sugges-tive of trabecular microfracture or a stress reaction; suchcases usually resolve more quickly and allow recom-mencement of graded physical activity within 2–3weeks and running at 6 weeks. Atypical, eccentric, non-linear stress reactions can be encountered where exces-sive loading is tractional at tendon insertions (Fig 20).

CT, although good for depicting overt fractures (Fig21), is less capable of depicting stress reactions withouta fracture line and does not provide adequate informa-tion to assess other potential differential diagnoses re-lating to soft-tissue injury. Bone scintigraphy has goodsensitivity for identifying stress injuries, but unless thefindings are bilateral and symmetric, scintigraphic find-ings are often nonspecific and cannot be used to dis-tinguish an early stress reaction from overt fracture, apoint that can influence substantially the planning of arehabilitation program versus surgical intervention.

Insufficiency Fractures

Insufficiency fractures occur when normal loads areapplied to a structurally deficient pelvic ring. Thesefractures are frequently multiple and characteristicallyare bilateral and relatively symmetric (Fig 22). MRimaging is the investigation of choice in elderly patientswith pelvic pain who are suspected of having insuffi-ciency fractures because the specificity is high whensymmetric abnormality is identified, even in the ab-sence of a discrete fracture line. CT is an alternative toMR imaging and can also be used when the findings at

Figure 16. Calcific myone-crosis in a 29-year-old drugabuser who had pain andswelling in lower portion ofleg with no acute inflamma-tory changes. History oftrauma to lower part of legin a motor vehicle accident8 years earlier was elicited.(a) Conventional radiographand (b) transverse CT imageshow extensive sheetlikecalcification throughout tibi-alis anterior compartment,indicative of calcificmyonecrosis.

Figure 17. Stress reaction in a 24-year-old marine with painfulright hip. Transverse STIR MR image shows marrow edema inposterior column of right hip, which, in the absence of joint effu-sion or inflammatory markers, was considered to representtraining-related stress reaction because no discrete fracture linewas identified. Training recommenced at 6 weeks.

Hu

gh

es

204

MR imaging are equivocal, particularly in the presenceof known malignancy, because CT excludes destructivechange that is usually seen in pathologic fractures, butCT is generally less sensitive than MR imaging. Bonescintigraphy is also a useful screening test, but casesdemonstrating asymmetry often require investigationwith either CT or MR imaging.

Insufficiency fractures commonly occur because ofosteoporosis, but other metabolic disorders predispos-ing to fracture include osteomalacia and parathyroid-related bone disease. In patients receiving radiationtherapy for gynecologic malignancy, Blomlie et al (9)identified insufficiency fractures occurring in as manyas 79% of the patients within 2 years of treatment.The MR imaging features either resolved or improvedwithin the 30-month duration of the study. Abe et al(10) also found insufficiency fractures, but these oc-

Figure 18. Sacral stress fracture in a 41-year-old distance jogger. (a) Coronal STIR MR image shows right-sided sacral edema andnormal joint. (b) Axial T1-weighted SE MR image shows incomplete fracture line. Training was successfully recommenced graduallyat 4 weeks.

Figure 21. Insufficiency fractures in a 71-year-old woman withlow back, sacral, and hip pain. CT image shows typical insuffi-ciency fractures (arrows) in both iliac bones.

Figure 20. Atypical stress reaction in a 40-year-old powerwalker with groin and left hip pain. Coronal STIR MR imageshows focal edema around left lesser trochanter (arrow), whichwas considered to represent less typical pattern of stress reac-tion. Symptoms resolved rapidly with rest.

Figure 19. Clini-cal presumptionof pubic sym-physitis in a 22-year-old soccerplayer. CoronalT2-weighted fastSE MR imageshows para-symphysealstress fracture(black arrow) andedema at adduc-tor insertions(white arrow).

Low

-Energ

y Inju

ries of th

e Low

er Limb

205

Conventional radiographic evaluation is usually ad-equate to establish the diagnosis, but diagnostic diffi-culty can be encountered in the skeletally immatureindividual in whom ossification at the origins of thesemuscles is limited. Both MR imaging and US can beused to establish a positive diagnosis in these cases,but the US option is dependent on there being localmusculoskeletal US expertise. US is usually immedi-ately available and well tolerated by young children(Fig 25), but MR imaging is often preferred because itprovides a more comprehensive evaluation in relationto more subtle muscle injuries or occult fractures inand around the pelvis, which are part of the workingdifferential diagnosis in such cases. US is generally lessaccurate around the pelvis because limited access andincreased depth of injury result in reduced imagequality and sensitivity.

Greater trochanteric avulsions are more commonlyencountered in elderly patients and require further in-vestigation. Associated intertrochanteric fractures arereported in as many as 66% of the patients, and mostof these fractures are occult and are best demonstratedby MR imaging (11,12). Intertrochanteric injuries thatextend across the femoral neck at MR imaging maywarrant fixation of the femoral neck with nailing toavoid subsequent overt fracture (Fig 25).

Long-standing and maturing avulsions may mani-fest as either (a) hypertrophic ossification simulatinga mass lesion or (b) localized erosion caused by hy-peremic osteolysis, which may be interpreted in erroras erosion by an adjacent mass lesion. In both cases,the site of the lesion should suggest the diagnosis,and MR imaging can be used to exclude a mass le-sion (Fig 26). MR imaging can also be used to identifycoexistent pathologic findings that can contribute to

Figure 24. Avulsion injury in a 25-year-old athlete with acutehip pain. Conventional radiograph shows avulsion by sartoriusof anterior superior iliac spine (arrow).

Figure 23. Radiograph of the three common sites of pelvicavulsion injuries: the anterior superior iliac spine (origin of sar-torius) (arrowhead), the anterior inferior iliac spine (origin ofrectus femoris) (long arrow), and the ischial tuberosity (originof hamstrings) (short arrow).Figure 22. Insufficiency fractures in a 65-year-old woman with

unexplained low back and pelvic pain. Coronal STIR MR imageshows bilateral symmetric areas of marrow edema in sacral alaeand ischial regions (arrows), typical of insufficiency fractures.

curred in only 34% of the patients following pelvic ir-radiation and in 84% of the patients with pelvic pain.Multiple fractures were found in 85%, and fractureswere symmetric in 67%. Unlike the insufficiency frac-tures, metastases, which were found in 3% of the pa-tients, were all outside the radiation field.

Avulsion Injuries

Avulsion injuries of the pelvic ring usually occur inyoung or skeletally immature individuals, commonlyathletes. The injuries follow isometric muscle contrac-tion and affect three main sites (Fig 23): the anteriorsuperior iliac spine (origin of sartorius) (Fig 24), theanterior inferior iliac spine (origin of rectus femoris),and the ischial tuberosity (origin of the hamstrings).

Hu

gh

es

206

sufficient to exclude injury, but additional coronaland axial T1-weighted SE and STIR MR images are re-quired to fully evaluate abnormalities identified withthe screening pulse sequence (Fig 28).

Groin Pain

Pain in the anterior portion of the pelvis and groinis a common manifestation of pelvic injury. In theUnited Kingdom, most injuries are encountered insportsmen and particularly soccer players, becausemany of these injuries occur as a result of twisting,turning, and kicking movements across the line of thebody. Symptoms and signs may guide clinical decision

Figure 25. Hamstring apophy-seal avulsion in a 12-year-oldboy. Sagittal US images of ham-string origin show normal leftside (L), with cortical line (whitearrow) capped with cartilaginousgrowth zone, and right side (R)with cortical avulsion (black ar-row) with surrounding hypoecho-ic hematoma.

Figure 26. Greater trochanteric avulsion in a 68-year-oldman following a minor fall. (a) Pelvic radiograph shows greatertrochanteric avulsion (arrow). On strength of conventional ra-diograph, MR imaging was suggested. (b) Coronal STIR MRimage shows intertrochanteric edema (arrows), but (c) coronalT1-weighted SE MR image demonstrates incomplete fractureline (arrows). This patient was managed nonoperatively andwas allowed to partially weight-bear at 4 weeks.

symptoms in avulsion injuries, such as the associationof sciatic neuritis with injury of the ischial tuberosity(Fig 27).

Occult Fractures and Soft-Tissue Injury

As indicated in the previous paragraphs, MR imag-ing provides the most efficient and accurate methodof diagnosing insufficiency fractures, but MR imagingshould also be considered early in the work-up of pa-tients with acute hip and pelvic pain following rela-tively low-energy falls in whom conventional radio-graphs are normal. Bone density need not necessarilybe abnormal, and injuries to structures other than thepelvic ring are often encountered, including the femo-ral neck and surrounding soft tissues, in these pa-tients. Bogost et al (13) showed pelvic ring fractures in23%, femoral neck fractures in 37%, and soft-tissueinjury in 74% of the patients with normal pelvic ra-diographs who were suspected of having occult injury.Screening with coronal STIR MR imaging is usually

Low

-Energ

y Inju

ries of th

e Low

er Limb

207

Figure 27. Repetitive tractional injury of left ischial tuberosity in a 12-year-old boy. (a) Anteroposterior radiograph shows bone re-sorption (arrows). (b) Coronal STIR MR image shows granulating hyperemic interface (arrowheads).

making, but the differential diagnosis is broad and in-cludes adductor injuries, hernias, pubic symphysitis,and stress fractures of the pubic rami.

The investigative pathway is influenced by the mostlikely clinical diagnosis. If a hernia is most likely, USis preferred. MR imaging is substantially less capableof providing information to establish the diagnosisof hernia, given that nearly all hernias are reducible,and the posterior wall defects are not identifiablewith nonstraining MR imaging. MR imaging is pre-ferred if the other diagnoses outlined previously aremore probable than a hernia. These patients shouldbe encouraged to exercise or stress the injury in the48 hours preceding MR imaging because subtle tearsor sprains will be identified only by the presence ofedema (Fig 29).

Overt inguinal hernias are well recognized in middle-aged nonathletic adults but are now increasingly di-agnosed as a cause of performance-limiting injury insports and athletics. The spectrum of disorder rangesfrom small tears in the transversalis fascia (sportsmanhernia) and posterior wall weakness without hernia-tion to direct or indirect herniation. Although a smallnumber of hernias are indirect, most are direct and areacquired through injury to the posterior wall of the me-dial end of the canal in the region of the external in-guinal ring. US of the inguinal canal is best achieved intransverse section, revealing an oval structure includingthe testicular artery and vein that can be identified withcolor flow imaging. The posterior wall is seen as a lin-ear reflective line that moves away from the probe withstraining. Weakness of the posterior wall without a de-fect or hernia is manifest by the posterior wall bulgingtoward the probe. A hernia is more focal and has moreangular interface with the posterior wall and can bemassaged back through the defect with gentle pressureapplied through the probe head. Most hernias containfat, but bowel is occasionally identified.

Figure 28. Hamstringavulsion injury in a 27-year-old man. (a) Transverse CTimage shows ischial avul-sion (arrows). (b) Axial T1-weighted SE and (c) axialSTIR MR images show as-sociated sciatic neuritis (ar-row), which caused severeradiating leg pain.

Hu

gh

es

208

In patients in whom a hernia is not the most likelydiagnosis, MR imaging is the preferred investigation.Coronal and axial T1-weighted SE and STIR (or T2-weighted fast SE with fat saturation) MR images pro-vide a good screening series but must include theproximal thighs and adductors. Both sides should beincluded to allow for comparison. T2-weighted fastSE MR images without fat saturation should not beused in preference to a fat-saturated pulse sequencebecause they have reduced sensitivity for musclesprains and marrow edema.

Pubic symphysitis is a stress-related reaction com-prising degeneration and herniation of the symphys-eal disk and reactive edema in the parasymphysealbone. Determining clinical significance is difficult be-cause this abnormality occurs commonly in otherwiseasymptomatic sportsmen, particularly soccer players.Gross asymmetry in parasymphyseal marrow edema,although not uncommon, can indicate stress fractureof the pubic rami, rather than primary symphysitis(Fig 19). Injection of long-acting local anestheticdrugs and steroids into the symphysis can clarify theclinical significance of symphyseal abnormality.

THIGH

Adductor Injuries

Tears and strains of the adductor musculature arecommon (Fig 30), and this location is one of themore frequent sites to encounter myositis ossificans.Chronic or subacute adductor insertion avulsion inju-ries are also encountered and manifest clinically as“thigh splints.” These injuries have been recorded inadults (14) and, more recently, in the pediatric popu-lation (15). The manifestation comprises exercise-related pain in the medial side of the thigh associatedwith periosteal reaction, periosteal edema at MR imag-ing, or increased radiotracer uptake at bone scintigra-phy in children. There are no associated mass lesions.

Figure 30. Adductor brevis injury in a 26-year-old squash player with groin pain. Previous MR images had shown pubic symphysitis,but when the patient was imaged within 24 hours of exercise, (a) transverse and (b) coronal MR images demonstrate left-sided grade1 adductor brevis injury (arrow), in addition to pubic symphysitis (arrowheads).

Figure 29. In-complete stressfracture or reac-tion in a 26-year-old marine withleft hip pain exac-erbated by exer-cise. CoronalSTIR MR imageshows incompletestress fracture orreaction acrossfemoral neck.

Knowledge of this entity is important because it canavoid the need for biopsy. Similar tractional periosti-tis can also be encountered at the insertion of thevastus medialis and vastus intermedius (15).

Hamstrings

The hamstring unit is the most frequently injuredmuscle group in sportsmen. The semimembranosusand biceps femoris are most commonly affected. Theprognosis in these patients relates to the magnitudeof cross-sectional involvement of the muscle affectedand is unrelated to the presence of fluid collections,areas of hemorrhage, or distal injury (16). As with alltears, identification of the muscle injured is primar-ily achieved by examination of the axial T1-weightedSE and STIR MR images, but unlike adductor tears, inwhich coronal images are supportive of axial data sets,

Low

-Energ

y Inju

ries of th

e Low

er Limb

209

pair. Because a large number of the referrals with sus-pected quadriceps rupture are erroneous, we usuallyscreen with US and reserve MR imaging for cases inwhich US cannot distinguish a high-grade partial tearfrom complete rupture, or if internal derangement ofthe knee is suspected. Without the option or expertiseto offer a US examination, it would be appropriate todefault to MR imaging as the primary investigation.Secondary features of quadriceps rupture includewaviness of the infrapatellar tendon, but this is a non-specific sign and may also be found following kneeinjuries due to inhibition of quadriceps muscles.

KNEE

Avulsion and Osteochondral Injuries

Formal review of MR imaging of the knee is beyondthe scope of this chapter, but several signs of ligamen-tous and osteochondral injury can be inferred fromconventional radiographic evaluation of the injuredknee, which should alert the reporting radiologist andclinician to associated injury.

Arcuate Sign or Fracture

Avulsion of the fibular collateral ligament insertioncan be identified on anteroposterior and lateral views,providing the head of the fibula is not superimposedon the tibia. This injury is often associated with arcu-ate ligament, popliteus tendon, and anterior cruciateligament disruption, and commonly requires surgicalintervention (18). Bone bruising is often an accompa-niment (Fig 32).

Segond Fracture

Lateral capsular avulsions extract a small fragmentof bone from the tibia posterior to the attachment ofthe iliotibial tract at Gerdy’s tubercle. Segond fracturesare associated with anterior cruciate ligament tears inmore than 90% of the patients (Fig 33).

Gerdy’s Avulsion

Injury to Gerdy’s tubercle results from a varus in-jury; the tubercle is avulsed by the iliotibial tract. Thedetached fragment is usually much larger than theSegond fracture fragment and is associated with lateralcapsular, fibular collateral, and anterior cruciate liga-ment injury (Fig 34).

Osteochondral Fracture

Shearing injuries at the osteochondral junction notinfrequently follow minor trauma. The cartilage defectsmay be substantial; nonetheless, many are not identifi-able on conventional radiographs. A thin layer of bonemay be avulsed across the base of the hyaline cartilage,and the identification of a linear intraarticular flake canusually be interpreted to represent a substantial osteo-chondral injury. MR imaging is advisable to determine

Figure 32. Avulsion injuries in a 15-year-old soccer player in-jured in a tackle. Anteroposterior radiograph shows avulsion offibular head (arcuate sign) (long arrow). In addition, secondlarge lateral fragment (short arrow) and smaller intercondylarfragment (arrowhead) are suggestive of avulsions of iliotibialtract and anterior cruciate ligament, respectively.

Figure 31. Recurrent hamstring tears in a 24-year-old soccerplayer who had recent severe pain radiating into lower portion ofleg. Axial STIR MR image shows semimembranosus tear, withedema surrounding sciatic nerve (arrow).

sagittal T1-weighted SE and STIR MR images are pre-ferred in a patient who is suspected of having ham-string injury. Tears occur in either the distal tendon orat the muscle-tendon junction anywhere along itslength (17). Patients with chronic and sometimes acutehamstring injury can present with sciatica caused by ir-ritation and inflammation of the sciatic nerve, whichshould be scrutinized on the axial images (Fig 31).

Quadriceps and Infrapatellar Tendons

Injuries to the quadriceps and infrapatellar tendons,as opposed to the muscle units, are uncommon andare as frequently seen in athletic as nonathletic indi-viduals. The diagnosis of infrapatellar tendon ruptureis easily made at US, but the complexity of the multi-layered quadriceps mechanism proximal to the patellais often challenging with either MR imaging or US.The main clinical question is to distinguish a partialfrom a complete tear, the latter requiring surgical re-

Hu

gh

es

210

its true size and origin. The patella is most frequentlyaffected, with the medial facet of the patella beingswept of its cartilage during lateral dislocation (Fig 35).Personal review of attempts to reattach such osteochon-dral fragments would suggest that these are generallyunsuccessful.

LOWER PART OF LEG

Stress Fractures and “Shin Splints”

Stress fractures in the lower limb are most commonin the tibia and foot. Although the injury may merelyreflect repetitive overloading in a biomechanically nor-mal limb, gait disorders and footwear can predisposeto injury. Most stress fractures occur in runners at thejunction of the mid and proximal thirds of the tibia.The earliest conventional radiographic sign is loss ofdefinition of the posterior cortex, followed by periosti-tis (Fig 36) and finally a linear zone of sclerosis.

Schweitzer and White (19) found that hyperpro-nation of the foot could elicit stress-related MR imag-ing changes in the bone marrow of the foot and, to alesser extent, the tibia and femur. In our practice, armedforces personnel not infrequently show changes con-sistent with stress-related injuries related to alter-ations in standard-issue footwear. Sagittal and axialimages are useful in distinguishing stress fracturesfrom nutrient foramen. The axial image also can beused to distinguish longitudinal stress fractures, withcortical clefts on axial images, from the more com-mon transverse fracture pattern (20). STIR images arethe most sensitive MR images in the identification ofstress fractures (21).

Stress-related injury in the tibia may also manifestwith a distinct clinical entity commonly called shin

splints. These patients have normal findings on con-ventional radiographs but can have MR imagingfindings ranging from normalcy to stress fractures.Other patients will have periosteal fluid and diffusemarrow change, usually along the anteromedial bor-der of the midportion of the tibia. Chronic shin painis invariably associated with normal findings at im-aging, indicating that MR imaging has less utility inthis group (22).

Achilles Injury

The Achilles tendon, in reality, is best considered asa muscle-tendon unit that comprises the tendon it-

Figure 33. Segond fracture in a 16-year-old male adolescent with twisting injury of knee. (a) Radiograph shows Segond fracturefragment (arrow) along lateral joint line. (b) Coronal T1-weighted SE MR image shows loss of definition of anterior cruciate ligament(black arrow) indicative of tear. The Segond fragment (white arrow) and donor site (arrowhead) are not easily appreciated.

Figure 34. Gerdy’s avulsion in a 20-year-old man injured in arugby tackle. Coronal proton-density–weighted (PD) MR imagewith fat saturation demonstrates iliotibial tract avulsion (black ar-row). Bone fragment is larger than Segond fragment and arisesfrom Gerdy’s tubercle in line with anterior meniscus, which is un-stable in this case (black arrowhead). Medial collateral injury(white arrow) and bone bruising (white arrowheads) are evident.

Low

-Energ

y Inju

ries of th

e Low

er Limb

211

self, the gastrocnemius and soleus muscles, and theircommon muscle-tendon junction, which extendshigh into the calf (Fig 37). Injuries can be encoun-tered at many sites, including the origins of the gas-trocnemius and the Achilles tendon itself, but mostinjuries affect the muscle-tendon junction along itsmedial side (Figs 38, 39).

Achilles Tendinopathy

Tendinopathy is the most common manifestationof chronic Achilles tendon overload, most frequently

manifests in middle age, and is more prevalent in flat-footed and obese individuals. Histopathologic analy-sis of these tendons reveals mucoid degeneration,with varying degrees of neovascularization. MR imag-ing is unnecessary in these patients because US bettershows the distribution of tendon degeneration (Fig40) and neovascularization (Fig 41). As a practicalmatter, US of the Achilles tendon can be masteredwith limited experience. Neovascularization is linkedto the symptoms but does not reflect a poor ultimateoutcome (23).

Figure 36. Periostitis and stress fracture ina 20-year-old marine recruit with pain inlower portion of leg. (a) Lateral radiographobtained at 5 days shows normalcy of initialconventional radiograph. (b) Lateral radio-graph at 15 days shows subsequent evolu-tion of periosteal reaction (arrows) consis-tent with stress fracture.

Figure 35. Osteochondral fracture in a 22-year-old woman with twisted knee and laterally dislocated patella. (a) Anteroposterior ra-diograph shows ossific fragments (arrow) in lateral joint recess. (b) Coronal PD fat-saturated MR image confirms intraarticular loosebody (arrows). (c) Axial PD fat-saturated MR image shows normal hyaline cartilage on lateral facet (long black arrow), but absentcartilage medially (short black arrow) and associated bone defect (arrowhead). Impaction bruising (white arrow) confirms lateral pa-tellar dislocation.

Hu

gh

es

212

Achilles Tendon Tears

The main clinical dilemma in relation to Achilles ten-don tears is to distinguish the following: (a) ruptured ten-don from ruptured muscle-tendon junction, (b) the posi-tion and size of the tendon deficit, (c) the ability to op-pose ruptured tendon ends with the foot plantar flexed,(d) partial from complete tear, and (e) tendinopathyfrom interstitial tear. Although MR imaging can performwell in relation to many of these questions, US is moreaccurate in relation to identifying tendon ends and thesize of defects and can also be used to evaluate whethertendon ends can be brought within 1 cm of each otherduring plantar flexion, which allows an equinus plas-ter to be applied as an alternative to surgical interven-tion. US also takes a fraction of the time that MR imag-ing takes and can be performed while the patient is in theemergency department. Both MR imaging and US shouldbe extended proximally to include the muscle bellies andmuscle-tendon junction in the midportion of the calf.

ANKLE

Ankle injuries are one of the most common causes forpatients to come to the hospital emergency depart-ment. While most injuries result from inversion (supi-nation) of the foot, eversion (pronation) and associ-ated rotation can result in specific patterns of ligamen-tous disruption and fracture. Although such injuriesare not associated with high-energy impaction or

Figure 40. Mucoid degeneration of right Achilles tendon in a35-year-old man. Sagittal US images of both Achilles tendonsshow superficial (long arrows) and deep (short arrows) aspectsof tendons. The right tendon is expanded with a hypoechoicventral aspect (∗) indicative of mucoid degeneration.

Figure 39.Grade 2 injuryof medial sideof muscle-tendon junc-tion of soleusin a 44-year-old man.Sagittal USimage dem-onstratesmuscle fiberdisruption attendon junc-tion (arrows).

Figure 37.Normalmuscle-tendon junc-tion on medialside of calf ina 24-year-oldman. US im-age showsgastrocne-mius and so-leus feedinginto commontendon, whichon high-reso-lution US im-ages oftenhas a bilay-ered appear-ance (arrows).

Figure 38. Grade 1 injury of medial sideof muscle-tendon junction of right gastroc-nemius in a 41-year-old man. Sagittal USimages show normal left side (LT) and onright side (RT) demonstrate edema or hem-orrhage separating muscle fibers (arrows)but little fiber disruption.

Low

-Energ

y Inju

ries of th

e Low

er Limb

213

rapid deceleration, it could be argued that the forcesapplied through focused redirection of the bodyweight across the mortise does in fact represent ahigh-energy injury. The spectrum of abnormality in-cludes ligament strains and tears, tibia and fibula frac-tures, talar dome injuries, tendon injury, and sinustarsi syndrome.

Fractures

Appropriate use of conventional radiographs re-mains the primary investigation of an ankle injury.Application of the Ottawa ankle rules (24) can al-low safe triage and onward referral for imagingwithout unnecessary x-ray exposures by either medi-cal or nursing staff. The optimal imaging protocolvaries, but an anteroposterior ankle view is standard,while institutions vary in their use of a true lateral viewand/or a mortise view. Marginal benefits have been

shown in the identification of medial and lateral mal-leolar fractures by combining all three views (25).

The important point in interpretation is to appre-ciate the mechanism and true extent of the injuryfrom the radiographic appearances, rather thanmerely reporting the presence of fractures. This willallow a far greater appreciation of associated liga-mentous injury and is facilitated by an understand-ing of the Lauge-Hansen classification of ankle frac-tures. This classification identifies specific patternsof lateral malleolar fracture that help to categorizefractures into one of four main patterns (Fig 42).Within each pattern, there is a constant order ofstructural failure that includes the associated liga-mentous structures. This system allows ligamentousinjuries to be inferred in certain circumstances andalso forms a framework for systematic analysis ofthe radiographs (Fig 43). Regardless of the viewsemployed or the technical expertise of the radiolo-gist, approximately 5% of the fractures cannot beidentified. In many of these patients, the injury ismanaged clinically as a fracture, but cross-sectionalimaging can be a great advantage to the diagnosis ofoccult bone and soft-tissue injury in these patients.

MR imaging can be used to identify occult medialand lateral malleolar fractures, osteochondral injury,and bone bruising of the talar dome (Fig 44). Ligamen-tous and tendon injuries and sinus tarsi inflammationcan also be identified. Because most of these patientsare non–weight bearing and because all would haveplaster (backslab) cast applied, MR imaging is not usu-ally considered until the patient is reviewed at a fractureclinic at 7–10 days, assuming the diagnosis remainsunclear from the repeat radiographs.

Figure 41. Hyperemic degenerative Achilles tendon in a 35-year-old male distance runner. Sagittal US image obtainedwith color Doppler flow imaging demonstrates superficialhypoechoic tendon degeneration (arrow) and associated hyper-emia (arrowheads).

Figure 43. Pro-nation–lateral rota-tion injury type 4.Radiograph showsposterior malleolarfracture (arrow).High fibular frac-ture is not in-cluded on this im-age (additional im-ages were re-quired). Deltoidand distal tibiofibu-lar injuries arelikely to coexist.

Figure 42. Pronation–lat-eral rotation type 3 injury.Radiograph shows high spi-ral fracture (arrow) typical ofthis pattern of fracture. Del-toid and distal tibiofibularligaments should be sus-pected in this situation, asthey would usually precedehigh fibular fracture.

Hu

gh

es

214

Ligament Injury

Given the commonly associated effusions, acuteligamentous injury can often be assessed adequatelywith conventional MR imaging. Cases of long-stand-ing instability without an effusion, however, are bet-ter assessed with MR arthrography, which also hassuperior sensitivity for subtle chondral defects (Fig45) or impingement lesions (Fig 46), which are of-ten included as part of the differential diagnosis atreferral. US, although capable of depicting the ante-rior talofibular ligament, does not allow a completeevaluation of all ligaments and is not capable of ad-dressing the alternative or coexistent pathologicfindings outlined previously and is not, therefore, ofprimary use in this circumstance.

Sinus Tarsi Abnormality

The sinus tarsi syndrome refers to the presence of aninflammatory process in the sinus tarsi following aninversion (supination) injury. Patients are initiallyconsidered to have sustained lateral ligament injury oroccult fracture but are referred for investigation whentheir condition fails to respond to standard immobili-zation. Clinical examination may identify pain duringmovement at the subtalar articulation. MR imaging inthe acute setting identifies an inflammatory mass inthe sinus tarsi that usually results from injury to theinterosseous ligaments and consequently leads tosubtalar instability. In later imaging of this disorder,associated degenerative change will also be identifiedin the posterior facets of the subtalar joint (Fig 47).

Tendon Injury

Evaluation of a specific tendon should be performedwith US because it is quick and comprehensive, allowscomparison with the contralateral side, and is unaf-fected by changing tendon orientation and associatedMR artifacts around the malleoli. MR imaging is an al-ternative if US expertise is unavailable or if the differen-tial diagnosis is broader and includes structures such asthe spring ligament or osseous abnormality that are be-yond the scope of US.

In summary, this chapter covers some of the low-energy or repetitive-strain injuries encountered in thelower limb. Although conventional radiographs havean important role in these injuries, early diagnosis ofpreradiographic bone abnormality and soft-tissue in-

Figure 46. Soft-tissue impinge-ment in lateralankle gutter in a29-year-old sol-dier. TransverseT1-weighted SEMR arthrogram im-age with fat satu-ration demon-strates soft tissuesinterposed in lat-eral gutter (arrow).

Figure 45. Chon-dral injury of talardome in a 25-year-old marine. Coro-nal T1-weightedSE MR arthrogramimage demon-strates a chondralflap (arrow) pro-jecting intosuperolateral jointspace.

Figure 44. Bone bruising of talar dome following a fall in a 20-year-old man. Coronal STIR MR image demonstrates extensivehigh signal intensity (arrow) in superomedial aspect of left talus,consistent with posttraumatic bone bruising.

Low

-Energ

y Inju

ries of th

e Low

er Limb

215

jury can often be accelerated and a definitive diagnosisestablished by the appropriate use of MR and US imag-ing and occasionally CT and nuclear radiography.

References1. Takebayashi S, Takasawa H, Banzai Y, et al. Sonographic

findings in muscle strain injury: clinical and MR imaging cor-relation. J Ultrasound Med 1995; 14:899–905.

2. Jackson DW, Feagin JA. Quadriceps contusion in youngathletes. J Bone Joint Surg Am 1973; 55:95–105.

3. Rothwell AG. Quadriceps hematoma: a prospective clinicalstudy. Clin Orthop 1982; 171:97–103.

4. Ryan JB, Wheeler JH, Hopkinson WJ, et al. Quadricepscontusions. Am J Sports Med 1991; 19:299–304.

5. Peck RJ, Metreweli C. Early myositis ossificans: a newechographic sign. Clin Radiol 1988; 39:586–588.

6. Holobinko JN, Damron TA, Scerpella PR, Hojnowski L. Cal-cific myonecrosis: keys to early recognition. Skeletal Radiol2003; 32:35–40.

7. O’Keefe RJ, O’Connell JX, Temple HT, et al. Calcific myo-necrosis: a late sequela to compartment syndrome of theleg. Clin Orthop 1995; 318:205–213.

8. Major NM, Helms CA. Sacral stress fractures in long-dis-tance runners. AJR Am J Roentgenol 2000; 174:727–729.

9. Blomlie V, Rofstad EK, Talle K, Sundfor K, Winderen M,Lien HH. Incidence of radiation-induced insufficiency frac-tures of the female pelvis: evaluation with MR imaging. AJRAm J Roentgenol 1996; 167:1205–1210.

10. Abe H, Nakamura M, Takahashi S, Maruoka S, Ogawa Y,Sakamoto K. Radiation-induced insufficiency fractures ofthe pelvis: evaluation with 99mTc-methylene diphosphonatescintigraphy. AJR Am J Roentgenol 1992; 158:599–602.

11. Craig JG, Moed BR, Eyler WR, van Holsbeeck M. Fracturesof the greater trochanter: intertrochanteric extension shownby MR imaging. Skeletal Radiol 2000; 29:572–576.

12. Learch TJ, Pathria MN. Greater trochanteric fractures: MRassessment and its influence on patient management.Emerg Radiol 2000; 7:89–92.

13. Bogost GA, Lizerbram EK, Crues JV III. MR imaging inevaluation of suspected hip fracture: frequency of unsus-pected bone and soft-tissue injury. Radiology 1995;197:263–267.

14. Charkes ND, Siddhivarn N, Schneck CD. Bone scanning inthe adductor insertion avulsion syndrome ("thigh splints"). JNucl Med 1987; 28:1835–1838.

15. Anderson SE, Johnston JO, O’Donnell R, Steinbach LS.MR imaging of sports-related pseudotumor in children: midfemoral diaphyseal periostitis at insertion site of adductormusculature. AJR Am J Roentgenol 2001; 176:1227–1231.

16. Slavotinek JP, Verrall GM, Fon GT. Hamstring injury in ath-letes: using MR imaging measurements to compare extentof muscle injury with amount of time lost from competition.AJR Am J Roentgenol 2002; 179:1621–1628.

17. De Smet AA, Best TM. MR imaging of the distribution andlocation of acute hamstring injuries in athletes. AJR Am JRoentgenol 2000; 174:393–399.

18. Juhng SK, Lee JK, Choi SS, Yoon KH, Roh BS, Won JJ.MR evaluation of the "arcuate" sign of posterolateral kneeinstability. AJR Am J Roentgenol 2002; 178:583–588.

19. Schweitzer ME, White LM. Does altered biomechanicscause marrow edema? Radiology 1996; 198:851–853.

20. Craig JG, Widman D, van Holsbeeck M. Longitudinal stressfracture: patterns of edema and the importance of the nutri-ent foramen. Skeletal Radiol 2003; 32:22–27.

21. Schmid MR, Hodler J, Vienne P, Binkert CA, Zanetti M.Bone marrow abnormalities of foot and ankle: STIR versusT1-weighted contrast-enhanced fat-suppressed spin-echoMR imaging. Radiology 2002; 224:463–469.

22. Anderson MW, Ugalde V, Batt M, Gacayan J. Shin splints:MR appearance in a preliminary study. Radiology 1997; 204:177–180.

23. Zanetti M, Metzdorf A, Kundert HP, et al. Achilles tendons:clinical relevance of neovascularization diagnosed withpower Doppler US. Radiology 2003; 227:556–560.

24. Verma S, Hamilton K, Hawkins HH, et al. Clinical applica-tion of the Ottawa ankle rules for the use of radiography inacute ankle injuries: an independent site assessment. AJRAm J Roentgenol 1997; 169:825–827.

25. Brandser EA, Berbaum KS, Dorfman DD, et al. Contributionof individual projections alone and in combination for radio-graphic detection of ankle fractures. AJR Am J Roentgenol2000; 174:1691–1697.

Figure 47. Sinus tarsi syndrome and associated subtalar joint degeneration in a 45-year-old man. (a) Coronal STIR MR imagedemonstrates an edematous inflammatory mass in sinus tarsi (arrows). (b) Sagittal STIR MR image shows focal areas of edema andearly cyst formation in the posterior facet of the talus (arrows), indicative of advanced degenerative change.

NO

TES

217

Pediatric LowerExtremity Trauma1

The patterns of musculoskeletal injury in infants, children, and adolescents differ con-siderably from those that occur in adults. The immature skeleton is structurally andmechanically different and responds differently to stress. The types of injury in thelower extremity are similar to those that occur in other portions of the immature skel-eton, but unique types of injury exist in certain locations. A thorough understandingof the pathophysiology and imaging characteristics of the common and sometimessubtle injuries that occur in pediatric patients is crucial for prompt diagnosis and ap-propriate management. Despite well-established trauma protocols and improved im-aging techniques, delays in diagnosis of such injuries still occur (1). Most fractures areadequately evaluated with radiographs, but cartilaginous and ligamentous injuriesmay require advanced imaging, such as magnetic resonance (MR) imaging. Computedtomography (CT) is useful for surgical planning with complex injuries at the joints.Ultrasonography (US) has limited usefulness in the detection of fractures and liga-mentous injuries. This chapter illustrates the unique features of injuries in pediatricpatients and includes discussion of optimal imaging strategies.

GENERAL CONSIDERATIONS IN EVALUATING THE IMMATURE SKELETON

The skeletal structures of young patients remain structurally and biomechanically im-mature. Bone is more porous, and the ratio of bone mineral to osteoid is lower in thepediatric skeleton throughout childhood. This difference in mineral content results ingreater elasticity in immature bone, allowing the bone to recoil after traumatic forcesup to the elastic limit. Forces that exceed the elastic limit result in fractures, which rangefrom simple plastic deformation to complete displaced fractures. The cartilaginousphysis and primary substantia spongiosa of the long bones are relatively weak in com-parison with ossified bone. The physis is most susceptible to shearing forces, whichcommonly occur with twisting injuries or oblique trauma. The Salter-Harris classifica-tion of fractures is typically used to categorize epiphyseal-metaphyseal injuries. Physesare also present adjacent to various apophyses that exist in the pediatric pelvis, femur,and foot. Excessive force applied to the muscles attached to these ossification centerscan result in avulsion fractures and chronic stress injuries. Ligaments are relativelystrong when compared to the physes; thus, isolated ligamentous injuries are less com-mon in children. However, ligament injuries associated with physeal fractures may re-quire more aggressive treatment and should be recognized.

Susan D. John, MD

RSNA Categorical Course in Diagnostic Radiology: Emergency Radiology 2004; pp 217–226.

1From the Department of Radiology, MSB2.100, University of Texas Houston Medical School, 6431 Fannin, Houston, TX77030 (e-mail: susan.d.john@uth.tmc.edu).

Joh

n

218

FRACTURE TYPES

Epiphyseal-Metaphyseal Fractures

Fractures involving the growing ends of the longand short tubular bones constitute 15%–30% of allfractures in childhood (2,3). The common feature ofthis category of fracture is injury to the cartilaginousphysis, with separation of the epiphysis from themetaphysis. Associated fractures of the epiphysis ormetaphysis and displacement of the epiphysis deter-mine the prognosis and management of these injuries.

Several classification systems have been proposed,but the most widely used classification is that de-scribed by Salter and Harris (4) in 1963. The numberof categories has been expanded by other investiga-tors, such as Ogden (5), who included new categoriesof fractures that do not directly involve the physis but

can result in growth disturbance, and Shapiro (6),whose classification is based on epiphyseal and meta-physeal blood supply in an attempt to better predictwhich fractures will result in premature closure of thephysis. A more recent classification by Peterson (7)adds categories for (a) transverse compression meta-physeal fractures with extension to the physis and(b) fractures with a portion of the physis missing(usually open fractures or gunshot wounds), becauseof the high incidence of unfavorable outcomes withsuch fractures. None of these classification systemsinclude the Salter-Harris type V fracture, an isolatedcrush injury of the physis that is rare, if it occurs at all.Nevertheless, the Salter-Harris system continues to befavored by most physicians, probably because of itssimplicity and familiarity.

Figure 1. Salter-Harris type I fracture ofdistal tibia. (a–d) AP and lateral radio-graphs show (a, c) mild widening of distalphysis (arrow) of right tibia, as comparedwith (b, d) normal contralateral tibial physis(arrow). (e) AP radiograph obtained 2weeks later shows sclerosis along themetaphysis at the physis, with mild adja-cent periosteal reaction, indicating healing.

Pediatric Lo

wer Extrem

ity Traum

a

219

Detection of Salter-Harris injuries of the lower ex-tremities on radiographs can be challenging when dis-placement of the fragments is minimal. In addition towell-positioned anteroposterior (AP) and lateralviews, oblique views are often helpful for identifyingsubtle fracture lines and physeal widening. Salter-Har-ris type I fractures, which are manifest only as widen-ing of the physeal line (Fig 1), can be difficult to diag-nose with confidence without a comparison view ofthe contralateral normal extremity.

The most common sites of physeal injuries of thelower extremities are the distal portions of the tibia andfibula (25% of epiphyseal injuries) (8) and the phalan-ges of the toes. Physeal injuries of the distal tibia andfibula can be serious because of the potential for asym-metric growth arrest, which leads to leg shorteningand angular joint deformity. Inversion (supination)injuries of the ankle are the most common, and frac-tures caused by this mechanism have the highest inci-dence of complications. Distraction of the joint later-ally results in Salter-Harris type I and II fractures of thedistal fibula or cortical avulsion injuries of the fibularepiphysis. Small avulsion fragments must be differenti-ated from normal accessory ossicles that are commonin the foot and ankle. US can be used to identify sub-periosteal hematoma or swelling of the peroneus lon-gus tendon that may signify an occult Salter-Harris typeI fracture of the fibula (9). However, these injuries areoften treated conservatively, with good results, on thebasis of clinical evidence alone without elaborate imag-ing. Ankle inversion (supination) may also impact themedial malleolus on the hindfoot, usually causing aSalter-Harris type III or IV fracture of the tibia.

Eversion (pronation) injuries of the ankle oftencause severe disruption of the ankle mortise. A widevariety of fractures can occur, including medial malle-olar avulsions, distal fibular shaft fractures, and Salter-Harris type III fractures through the lateral tibial epiph-ysis. The distal tibia is the most common site forSalter-Harris type III fractures, most often occurringwhen the physis is partially fused between the ages of12 and 15 years. Fusion of the medial aspect of thephysis precedes lateral fusion, and thus fractures tendto involve the lateral portion of the tibial epiphysis,the pediatric equivalent of a Tillaux fracture (Fig 2).

Another type of fracture that occurs when the tibialphysis is partially fused is the triplane fracture, acomplex injury resulting from a combination of axialloading and external rotation of the foot on the tibiawith the foot in plantar flexion. Fractures occur axiallythrough the unfused portion of the distal tibial physis,sagittally through the epiphysis, and coronally throughthe metaphysis (Fig 3). The fracture may consist of two,three, or four components, depending on the locationsof the fracture lines with respect to the partially fusedphysis. Optimal treatment requires internal fixation ifgreater than 2 mm of displacement of the fragments re-mains after attempted closed reduction. The complexanatomic structure of this fracture can be difficult to as-certain on radiographs, and CT is generally used to pro-vide a higher degree of accuracy and detail (10). MRimaging can also be used and may be more sensitivefor minimally displaced fractures.

Epiphyseal-metaphyseal fractures are less commonat the knee than at the ankle in children. Epiphysealseparations involve the distal femur more commonly

Figure 2. Salter-Harris type III fracture of distal tibia (Tillaux fracture). (a) AP radiograph shows that the displaced fragment (arrow)of the lateral epiphysis of the distal tibia is partially obscured by the overlying fibula. (b) AP radiograph from another patient showssubtle epiphyseal fracture (arrow). (c) Coronal reconstruction CT image of same patient as in b demonstrates more clearly the extentof the fracture and degree of displacement.

Joh

n

220

than the proximal tibia and are most often seen inadolescents with major trauma. When the fragmentsare displaced, the patient should be carefully examinedfor injuries to the popliteal artery or peroneal nerve.Salter-Harris injuries are common in the distal femurof older children and adolescents (Fig 4).

Physeal injuries at the knee in infants usually arecaused by birth trauma or child abuse. Nondisplacedfractures of the primary spongiosa paralleling the physisare common in child abuse and are virtually pathogno-monic in an infant with healthy bones. Although theseinjuries resemble Salter-Harris type II fractures, Kleinman

et al (11) revealed that the injuries are generally con-fined to the metaphysis and differ from epiphyseal sepa-rations that tend to occur with accidental trauma. Theclassic metaphyseal fractures of child abuse occur withviolent twisting of the lower extremities, either directlyor indirectly during episodes of forceful shaking. Smallcorner fragments or entire rims of the metaphysis maybe visible as separate fragments on radiographs (Fig 5).These fractures can be subtle and require high-resolutionwell-collimated images. Similar fractures can occur acci-dentally in infants with fragile bones caused by congeni-tal infection or metabolic bone disease.

Figure 4. Salter-Harris injury of the distal femur. (a, b) AP and lateral radiographs faintly show fracture (arrows) extending throughdistal femoral epiphysis. Only minimal irregularity is seen along the medial aspect of the distal femoral metaphysis. (c) Coronal pro-ton-density–weighted fat-saturated MR image more accurately defines the fractures through the epiphysis and metaphysis.

Figure 3. Triplane fracture ofdistal tibia. (a) AP view showsfractures transversely throughthe physis (arrows) and sagit-tally through the epiphysis (ar-rowhead). (b) Coronal meta-physeal fracture (arrow) is clearlyvisible only on the lateral view.

Pediatric Lo

wer Extrem

ity Traum

a

221

The proximal femoral physis is a relatively weakpoint at the hip and is a fairly common site of physealinjuries, especially in young children. Salter-Harristype I and II fractures are most frequent and are mani-fest as widening of the physeal line. In some patients,the metaphysis becomes displaced from the epiphysis,either acutely with trauma or on a subacute or chronicbasis with slipped capital femoral epiphysis. Whateverthe cause of the displacement, prompt immobiliza-tion and open reduction with internal fixation usuallyare warranted. Complications such as avascular necro-sis, varus deformity, early physeal closure, and non-union are common and can result in serious long-term disability (12).

Buckle (Torus) Fractures

Longitudinal (axial) force applied to long and shorttubular bones in children often results in failure in thethin cortex, which buckles along with the underlyingtrabecular bone (13). Buckle fractures are much morecommon in young children but can be seen through-out childhood and adolescence. Such fractures are ex-ceedingly rare in the mature skeleton. With bucklefractures, no radiolucent fracture line is visible, andcortical disruption is typically absent. This type offracture is most common in the metaphysis near thegrowth plate, where the cortex is thinnest, but buck-ling can also occur some distance away in the diaphy-sis. The milder forms of this type of fracture may showcortical buckling that involves the cortex along onlyone side of the bone (Fig 6). In some cases, the frac-ture causes only slight angulation in the normallystraight cortex. Care must be taken to identify subtlefracture extensions from the buckled cortex to thenearby physis, which may alter the treatment andprognosis of the injury.

Young children frequently sustain buckle-type frac-tures in the proximal and distal metaphyses of thetibia and fibula. Buckling of the anterior cortex of theproximal tibia may be mistaken for the yet undevel-oped tibial tuberosity if a comparison view of the nor-mal contralateral tibia is not available (Fig 7). Bucklefractures of the metatarsals are also common andshould be sought in young children who presentwith a limp and no history of known trauma. Suchfractures often result from a fall accompanied by im-paction of the forefoot against the ground. Bucklefractures of the first metatarsal are especially common

Figure 6. Buckle fracture of distal femur. (a, b) Radiographs show mild buckling of the distal femoral metaphysis (arrow) that ismore prominent anteriorly on (a) lateral view than medially on (b) AP view. (c) With healing, AP radiograph shows linear sclerosisthat reveals the complete transverse extent of buckle fracture (arrows).

Figure 5. Metaphyseal fractures of child abuse in a young in-fant. AP radiograph shows healing corner-type fractures (arrows),caused by nonaccidental trauma.

Joh

n

222

after a child jumps from a height, thus acquiring thenickname of “bunk-bed fractures” (Fig 8). The samemechanism can lead to impaction fractures of thecuboid when stress is greater along the lateral aspectof the foot. Cuboid fractures may be occult acutely,only appearing as linear sclerosis during healing (Fig9). Buckle and transverse fractures of the calcaneus arealso seen. Calcaneal fractures tend to be more benignin children than in adults, despite a fairly high inci-dence of involvement of the posterior subtalar joint(14). Like tarsal fractures, calcaneal fractures can bedifficult to detect when acute, but linear or archlikesclerosis develops as the fracture begins to heal (15).

Plastic Deformation Injuries and GreenstickFractures

Immature bone is more resilient under bendingforces because of greater elastic recoil, but when theelastic limit is reached, bowing deformity (plastic de-formation) remains after the force is released. Such in-juries are accompanied by microscopic infractionsalong the convex edge of the deformity; however, nofracture line is visible radiographically. The only evi-dence of plastic deformation on radiographs is the al-tered contour of the bone. Because the long bones ofthe extremities tend to have a normally curved contour,bowing fractures are difficult to detect unless they aresevere. Comparison with the uninjured extremity is aninvaluable aid in detecting these subtle injuries (16,17).The fibula is the most common bone in the lower ex-tremity to sustain a plastic deformation injury.

Greenstick fractures result from the same type ofbending force that causes plastic deformation, butwith greater intensity of force. Failure occurs through

the cortex along the convex surface of the deformityonly. In the leg, plastic deformation of one bone willoften accompany a greenstick fracture of the adjacentbone. Although the bowing injury appears relatively

Figure 8. "Bunk-bed" fracture of the first metatarsal. AP ra-diograph shows mild buckling (arrows) of the cortices of theproximal metaphysis of the first metatarsal, which indicates animpaction injury.

Figure 7. Buckle fracture ofthe proximal tibia. (a, b) Radio-graphs show slight cortical buck-ling (arrow) of the proximal tibia,which is slightly more prominenton (b) lateral view than on(a) AP view.

Pediatric Lo

wer Extrem

ity Traum

a

223

fracture. Therefore, plastic deformation injury must beidentified promptly and reduced if necessary. Angula-tion deformity of less than 20° will usually remodel inyoung children, but in children older than 4 years orwith a greater degree of angulation, correction of thedeformity should be considered.

Transverse, Oblique, and Spiral Fractures

Transverse, oblique, and spiral fractures are com-mon in the lower extremities in children and adoles-cents, and imaging evaluation does not differ substan-tially from that used in adults. However, the incidenceof fractures in certain locations is different in the im-mature skeleton. Femoral neck fractures are rare inotherwise healthy children and almost always arecaused by high-velocity trauma, fall from a height, orother severe types of trauma (18). Transcervical frac-tures account for from 20% to 50% of femoral neckfractures in children (19–21). The higher incidence offractures in this region is partially caused by the fre-quency of bone lesions, such as bone cysts and me-tastases, in the femoral neck. Careful assessment forsuch lesions should occur when a femoral neck frac-ture is found with no history of major trauma.

Fractures of the femoral shaft can result from a vari-ety of traumatic insults, including direct blows, rota-tional forces, and high-velocity axial loading. Femoralfractures in older children require forces of greatermagnitude, but young ambulating children can sus-tain spiral or oblique fractures with relatively minortwisting injuries. In infants, femoral shaft fractures areuncommon and should raise at least moderate suspi-cion of nonaccidental trauma (22,23). The majority offemoral shaft fractures occur in the midportion of theshaft. The degree of overriding of the fracture frag-ments on a radiograph obtained without tractionhelps to determine whether the fracture should bemanaged with early cast immobilization or traction. Amild degree of overriding (1–2 cm) is desirable inchildren between the ages of 2 and 10 years to preventovergrowth of the femur during healing. Supracondy-lar femur fractures tend to become anteriorly dis-placed because of contraction of the gastrocnemius,and thus they are usually managed with pin tractionor external fixation. Most femoral shaft fractures areisolated injuries, but care should be taken to excludean associated femoral neck or physeal fracture.

Transverse fractures are more common in the proxi-mal tibia than in the femur; and in younger children,these fractures can be hairline and subtle (Fig 10). Proxi-mal tibial fractures in children often result from impac-tion and hyperextension of the leg at the knee and maybe accompanied by buckling of the anterior cortex ofthe proximal tibia. A similar fracture can occur in olderchildren who are injured while jumping on a trampo-line. Valgus angulation at the fracture site duringhealing of these fractures is not uncommon. Relatively

Figure 9. Cuboid fracture. Radiograph shows linear sclerosis(arrow) without cortical buckling.

Figure 10.Hairlinetransversefracture ofproximaltibia.(a, b) AP andlateral radio-graphs showsubtle trans-verse radiolu-cency (ar-row). (b) Lat-eral viewshows mildanterior buck-ling at frac-ture site.

innocuous, the deformity often remodels slowly andmay interfere with adequate reduction of the adjacent

Joh

n

224

Figure 13. Avulsion fractures around the hips. (a) AP radiograph of pelvis showsthe subtle asymmetry (arrow) of the margins of the iliac bone just above the rightacetabulum. (b) Oblique view of the right hip more clearly demonstrates avulsion ofossification center (arrow) from anterior inferior iliac spine. (c) AP radiograph ob-tained from another child shows avulsion (arrow) of the lesser trochanter.

Figure 12.Stress fracture ofthe proximal tibia.Lateral radiographshows corticalthickening (arrow)along the dorsalaspect of theproximal tibial di-aphysis.

Figure 11. Tod-dler fracture of thetibia. Hairline spiralfracture (arrows) ofthe distal tibial di-aphysis was faintlyvisible only on APview.

minor twisting injuries of the ankle and leg can causenondisplaced spiral fractures of the tibial shaft (Fig 11).This fracture is so common in children between 1 and 2years of age that it has been called the “toddler fracture.”Hairline spiral tibial fractures can present a diagnosticchallenge because of their subtlety, the frequent lack of ahistory of trauma, and the tendency for manifestation inhealing states when the fracture line may not be easily

Pediatric Lo

wer Extrem

ity Traum

a

225

visible. Symptoms may mimic an ankle injury, andtibial radiographs should be considered whenever theclinical findings are suggestive of an ankle fracture in ayoung patient but none is found.

Avulsion and Stress Fractures

Stress fractures in children have the same imagingcharacteristics as those that occur in adults, but thedistribution of stress fracture sites differs slightly inyounger patients. Stress fractures are common in theproximal tibia (24), in the same location that hairlinetransverse and buckle-type fractures often occur. Asthey heal, cortical thickening becomes apparent alongthe dorsal aspect of the proximal tibia (Fig 12). Othercommon sites include the cuboid, fibula, and pelvis.As with adult stress fractures, the lesions are occult ra-diographically in the acute stage. MR imaging is sensi-tive for localizing acute stress injuries.

Avulsion fractures in the immature skeleton usuallyresult from excessive forces applied to ligaments thatattach to unfused apophyseal centers. These injuriesmay occur acutely, but they more commonly occursubacutely from chronic repetitive traction. The mul-tiple apophyses in the hips and pelvis fuse relativelylate, allowing ample time for avulsion injuries to oc-cur during adolescence. The separation of the apophy-seal ossification centers from the femur (lesser orgreater trochanter) or the pelvis (anterior superioriliac spine, anterior inferior iliac spine, ischium, iliaccrest) may be subtle and sometimes requires obliqueviews (Fig 13) or comparison views for verification.MR imaging can be used to depict these injuries, butthe findings can be less specific and may be suggestiveof other conditions, such as osteomyelitis, unless ra-

diographic confirmation of the fracture is made. Mostpelvic avulsion injuries heal with conservative man-agement. Avulsion fractures in the lower extremities,however, tend to be more unstable and require pinfixation. Acute avulsion of the tibial tubercle is a com-mon example of such an injury (Fig 14). Tibial tuber-osity avulsion fractures typically occur during sports,especially basketball (25). Acute tubercle avulsionsshould not be confused with the stable chronic formof injury at this site, Osgood-Schlatter disease.

In conclusion, the unique properties of the imma-ture skeleton are responsible for a relatively high inci-dence of extremity fractures that are subtle and diffi-cult to detect. The best imaging tools available forevaluating extremity fractures in infants and childrenare well-positioned and optimally exposed radio-graphs. Such images will reveal the fractures in almostall cases, particularly when the interpreter has a thor-ough understanding of the peculiarities of fractures inthe younger patient. CT and MR imaging can help todetect occult bone injuries and nonosseous structures.

References1. Furnival RA, Woodward GA, Schunk JE. Delayed diagnosis

of injury in pediatric trauma. Pediatrics 1996; 98:56–62.2. Mann DC, Rajmaisa S. Distribution of physeal and non-

physeal fractures in 2,650 long-bone fractures in childrenaged 0-16. J Pediatr Orthop 1990; 10:713–716.

3. Mizuta T, Benson WM, Foster BK, et al. Statistical analysisof the incidence of physeal injuries. J Pediatr Orthop 1987;7:518–523.

4. Salter RB, Harris WR. Injuries involving the epiphyseal plate.J Bone Joint Surg Am 1963; 45:587–622.

5. Ogden JA. Injury to the growth mechanism of the immatureskeleton. Skeletal Radiol 1981; 6:237–253.

6. Shapiro F. Epiphyseal growth plate fracture-separation: apathophysiologic approach. Orthopedics 1982; 5:720–736.

7. Peterson HA. Physeal fractures. III. Classification. JPediatr Orthop 1994; 14:439–448.

8. Devalentine SJ. Epiphyseal injuries of the foot and ankle.Clin Podiatr Med Surg 1987; 4:279–310.

9. Gleeson AP, Stuart MJ, Wilson B, Phillips B. Ultrasoundassessment and conservative management of inversion in-juries of the ankle in children: plaster of Paris versusTubigrip. J Bone Joint Surg Br 1996; 78:484–487.

10. Felman AH. Tillaux fractures of the tibia (in adolescents).Pediatr Radiol 1989; 20:87–89.

11. Kleinman PK, Marks SC, Blackbourne B. The metaphyseallesion in abused infants: a radiologic-histopathologic study.AJR Am J Roentgenol 1986; 146:895–905.

12. Rattey T, Piehl F, Wright JG. Acute slipped capital femoralepiphysis. J Bone Joint Surg Am 1996; 78:398–402.

13. Light TR, Ogden DA, Ogden JA. The anatomy of meta-physeal torus fractures. Clin Orthop 1984; 188:103–111.

14. Wiley JJ, Profitt A. Fractures of the os calcis in children.Clin Orthop 1984; 188:131–138.

15. Schindler A, Mason DF, Allington NJ. Occult fracture of thecalcaneus in toddlers. J Pediatr Orthop 1996; 16:201–205.

16. Borden S IV. Roentgen recognition of acute plastic bowingof the forearm in children. Am J Roentgenol Radium TherNucl Med 1975; 125:524–530.

17. Neumann L. Acute plastic bowing fractures of both the tibiaand the fibula in a child. Injury 1990; 21:122–123.

Figure 14. Avul-sion fracture of thetibial tuberosity.Lateral radiographshows completeseparation of theossification centerof the tibial tuber-osity (arrow).

Joh

n

226

18. Azouz EM, Karamitsos C, Reed MH, Baker L, Lozlowski K,Hoeffel JC. Types and complications of femoral neck frac-tures in children. Pediatr Radiol 1993; 23:415–420.

19. Morrissy R. Hip fractures in children. Clin Orthop 1980;152:202–210.

20. Swiontkowski MR, Winquist RA. Displaced hip fractures inchildren and adolescents. J Trauma 1986; 26:384–388.

21. Davison BL, Weinstein SL. Hip fractures in children: a long-term follow-up study. J Pediatr Orthop 1992; 12:355–358.

22. Gross RH, Stranger M. Causative factors responsible forfemoral fractures in infants and young children. J PediatrOrthop 1983; 3:341–343.

23. Dalton HJ, Slovis TM, Helfer RF, Comstock J, Schuerer S,Riolo S. Undiagnosed abuse in children younger than 3years with femoral fractures. Am J Dis Child 1990;144:875–878.

24. Ohta-Fukushima M, Mutoh Y, Takasugi S, Iwata H, Ishii S.Characteristics of stress fractures in young athletes under20 years. J Sports Med Phys Fitness 2002; 42:198–206.

25. McKoy BE, Stanitski CL. Acute tibial tubercle avulsion frac-tures. Orthop Clin North Am 2003; 34:397–403.